kashif HF staff commited on
Commit
9ce215a
1 Parent(s): c7181c2

initial monash time series forecasting repository (#3743)

Browse files

* initial monash time series forecast

* remove

* format

* initial 2 domains

* convert tsf to dataframe

* added some more config arguments

* the case when there is no item_id_column

* added weather and default start if not available

* added tourism dataset

* use Q-JAN freq

* added cif dataset

* added london smart meters

* added australian_electricity_demand

* added wind_farms_minutely

* added bitcoin

* added pedestrian counts

* vehicle trips dataset

* fix filename

* kdd_cup_2018 dataset

* fix record id

* added nn5 dataset

* web traffic and solar

* web traffic weekly

* cleanup

* added car parts

* added fred_md

* add traffic and rideshare

* formatting and make rideshare multivariate

* fix data column name

* added hospital and covid

* added single long time series

* fix freq

* initial readme

* fix typo

* added map for sec freq

* dataset info

* added dummy data

* more docs

* added multivariate forecasting task

* Update datasets/monash_tsf/README.md

Co-authored-by: Mario Šaško <mario@huggingface.co>

* Update datasets/monash_tsf/README.md

Co-authored-by: Mario Šaško <mario@huggingface.co>

* remove comments

* added data column

* added prediction lengths

* Fixed comment of the data split

* added missing reference

* added curators

* more curation information

* fix description

* move task to 2 sections

* make ROOT_URL a global

* updated description

* initial monash time series forecast

* remove

* format

* initial 2 domains

* convert tsf to dataframe

* added some more config arguments

* the case when there is no item_id_column

* added weather and default start if not available

* added tourism dataset

* use Q-JAN freq

* added cif dataset

* added london smart meters

* added australian_electricity_demand

* added wind_farms_minutely

* added bitcoin

* added pedestrian counts

* vehicle trips dataset

* fix filename

* kdd_cup_2018 dataset

* fix record id

* added nn5 dataset

* web traffic and solar

* web traffic weekly

* cleanup

* added car parts

* added fred_md

* add traffic and rideshare

* formatting and make rideshare multivariate

* fix data column name

* added hospital and covid

* added single long time series

* fix freq

* initial readme

* fix typo

* added map for sec freq

* dataset info

* added dummy data

* more docs

* added multivariate forecasting task

* Update datasets/monash_tsf/README.md

Co-authored-by: Mario Šaško <mario@huggingface.co>

* Update datasets/monash_tsf/README.md

Co-authored-by: Mario Šaško <mario@huggingface.co>

* remove comments

* added data column

* added prediction lengths

* Fixed comment of the data split

* added missing reference

* added curators

* more curation information

* fix description

* move task to 2 sections

* make ROOT_URL a global

* updated description

* added dataset usage section

* added annotation text

* guard for freq or prediction_length

* replace Exception with ValueErrors

* removed todo

* removed freq from Builder config

* remove all but weather dummy data

Co-authored-by: Shoaib Burq <saburq@gmail.com>
Co-authored-by: Mario Šaško <mario@huggingface.co>

Commit from https://github.com/huggingface/datasets/commit/d0237f4e552f055727f788066c033ed1e0987de2

Files changed (5) hide show
  1. README.md +234 -0
  2. dataset_infos.json +1 -0
  3. dummy/weather/1.0.0/dummy_data.zip +3 -0
  4. monash_tsf.py +566 -0
  5. utils.py +188 -0
README.md ADDED
@@ -0,0 +1,234 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ annotations_creators:
3
+ - no-annotation
4
+ language_creators:
5
+ - found
6
+ languages:
7
+ - unknown
8
+ licenses:
9
+ - cc-by-4-0
10
+ multilinguality:
11
+ - monolingual
12
+ pretty_name: Monash Time Series Forecasting Repository
13
+ size_categories:
14
+ - 1K<n<10K
15
+ source_datasets:
16
+ - original
17
+ task_categories:
18
+ - time-series-forecasting
19
+ task_ids:
20
+ - univariate-time-series-forecasting
21
+ - multivariate-time-series-forecasting
22
+ ---
23
+
24
+ # Dataset Card for Monash Time Series Forecasting Repository
25
+
26
+ ## Table of Contents
27
+ - [Table of Contents](#table-of-contents)
28
+ - [Dataset Description](#dataset-description)
29
+ - [Dataset Summary](#dataset-summary)
30
+ - [Supported Tasks and Leaderboards](#supported-tasks-and-leaderboards)
31
+ - [Languages](#languages)
32
+ - [Dataset Structure](#dataset-structure)
33
+ - [Data Instances](#data-instances)
34
+ - [Data Fields](#data-fields)
35
+ - [Data Splits](#data-splits)
36
+ - [Dataset Creation](#dataset-creation)
37
+ - [Curation Rationale](#curation-rationale)
38
+ - [Source Data](#source-data)
39
+ - [Annotations](#annotations)
40
+ - [Personal and Sensitive Information](#personal-and-sensitive-information)
41
+ - [Considerations for Using the Data](#considerations-for-using-the-data)
42
+ - [Social Impact of Dataset](#social-impact-of-dataset)
43
+ - [Discussion of Biases](#discussion-of-biases)
44
+ - [Other Known Limitations](#other-known-limitations)
45
+ - [Additional Information](#additional-information)
46
+ - [Dataset Curators](#dataset-curators)
47
+ - [Licensing Information](#licensing-information)
48
+ - [Citation Information](#citation-information)
49
+ - [Contributions](#contributions)
50
+
51
+ ## Dataset Description
52
+
53
+ - **Homepage:** [Monash Time Series Forecasting Repository](https://forecastingdata.org/)
54
+ - **Repository:** [Monash Time Series Forecasting Repository code repository](https://github.com/rakshitha123/TSForecasting)
55
+ - **Paper:** [Monash Time Series Forecasting Archive](https://openreview.net/pdf?id=wEc1mgAjU-)
56
+ - **Leaderboard:** [Baseline Results](https://forecastingdata.org/#results)
57
+ - **Point of Contact:** [Rakshitha Godahewa](mailto:rakshitha.godahewa@monash.edu)
58
+
59
+ ### Dataset Summary
60
+
61
+ The first comprehensive time series forecasting repository containing datasets of related time series to facilitate the evaluation of global forecasting models. All datasets are intended to use only for research purpose. Our repository contains 30 datasets including both publicly available time series datasets (in different formats) and datasets curated by us. Many datasets have different versions based on the frequency and the inclusion of missing values, making the total number of dataset variations to 58. Furthermore, it includes both real-world and competition time series datasets covering varied domains.
62
+
63
+ The following table shows a list of datasets available:
64
+
65
+ | Name | Domain | No. of series | Freq. | Pred. Len. | Source |
66
+ |-------------------------------|-----------|---------------|--------|------------|-------------------------------------------------------------------------------------------------------------------------------------|
67
+ | weather | Nature | 3010 | 1D | 30 | [Sparks et al., 2020](https://cran.r-project.org/web/packages/bomrang) |
68
+ | tourism_yearly | Tourism | 1311 | 1Y | 4 | [Athanasopoulos et al., 2011](https://doi.org/10.1016/j.ijforecast.2010.04.009) |
69
+ | tourism_quarterly | Tourism | 1311 | 1Q-JAN | 8 | [Athanasopoulos et al., 2011](https://doi.org/10.1016/j.ijforecast.2010.04.009) |
70
+ | tourism_monthly | Tourism | 1311 | 1M | 24 | [Athanasopoulos et al., 2011](https://doi.org/10.1016/j.ijforecast.2010.04.009) |
71
+ | cif_2016 | Banking | 72 | 1M | 12 | [Stepnicka and Burda, 2017](https://doi.org/10.1109/FUZZ-IEEE.2017.8015455) |
72
+ | london_smart_meters | Energy | 5560 | 30T | 60 | [Jean-Michel, 2019](https://www.kaggle.com/jeanmidev/smart-meters-in-london) |
73
+ | australian_electricity_demand | Energy | 5 | 30T | 60 | [Godahewa et al. 2021](https://openreview.net/pdf?id=wEc1mgAjU-) |
74
+ | wind_farms_minutely | Energy | 339 | 1T | 60 | [Godahewa et al. 2021](https://openreview.net/pdf?id=wEc1mgAjU- ) |
75
+ | bitcoin | Economic | 18 | 1D | 30 | [Godahewa et al. 2021](https://openreview.net/pdf?id=wEc1mgAjU- ) |
76
+ | pedestrian_counts | Transport | 66 | 1H | 48 | [City of Melbourne, 2020](https://data.melbourne.vic.gov.au/Transport/Pedestrian-Counting-System-Monthly-counts-per-hour/b2ak-trbp) |
77
+ | vehicle_trips | Transport | 329 | 1D | 30 | [fivethirtyeight, 2015](https://github.com/fivethirtyeight/uber-tlc-foil-response) |
78
+ | kdd_cup_2018 | Nature | 270 | 1H | 48 | [KDD Cup, 2018](https://www.kdd.org/kdd2018/kdd-cup) |
79
+ | nn5_daily | Banking | 111 | 1D | 56 | [Ben Taieb et al., 2012](https://doi.org/10.1016/j.eswa.2012.01.039) |
80
+ | nn5_weekly | Banking | 111 | 1W-MON | 8 | [Ben Taieb et al., 2012](https://doi.org/10.1016/j.eswa.2012.01.039) |
81
+ | kaggle_web_traffic | Web | 145063 | 1D | 59 | [Google, 2017](https://www.kaggle.com/c/web-traffic-time-series-forecasting) |
82
+ | kaggle_web_traffic_weekly | Web | 145063 | 1W-WED | 8 | [Google, 2017](https://www.kaggle.com/c/web-traffic-time-series-forecasting) |
83
+ | solar_10_minutes | Energy | 137 | 10T | 60 | [Solar, 2020](https://www.nrel.gov/grid/solar-power-data.html) |
84
+ | solar_weekly | Energy | 137 | 1W-SUN | 5 | [Solar, 2020](https://www.nrel.gov/grid/solar-power-data.html) |
85
+ | car_parts | Sales | 2674 | 1M | 12 | [Hyndman, 2015](https://cran.r-project.org/web/packages/expsmooth/) |
86
+ | fred_md | Economic | 107 | 1M | 12 | [McCracken and Ng, 2016](https://doi.org/10.1080/07350015.2015.1086655) |
87
+ | traffic_hourly | Transport | 862 | 1H | 48 | [Caltrans, 2020](http://pems.dot.ca.gov/) |
88
+ | traffic_weekly | Transport | 862 | 1W-WED | 8 | [Caltrans, 2020](http://pems.dot.ca.gov/) |
89
+ | hospital | Health | 767 | 1M | 12 | [Hyndman, 2015](https://cran.r-project.org/web/packages/expsmooth/) |
90
+ | covid_deaths | Health | 266 | 1D | 30 | [Johns Hopkins University, 2020](https://github.com/CSSEGISandData/COVID-19) |
91
+ | sunspot | Nature | 1 | 1D | 30 | [Sunspot, 2015](http://www.sidc.be/silso/newdataset) |
92
+ | saugeenday | Nature | 1 | 1D | 30 | [McLeod and Gweon, 2013](http://www.jenvstat.org/v04/i11) |
93
+ | us_births | Health | 1 | 1D | 30 | [Pruim et al., 2020](https://cran.r-project.org/web/packages/mosaicData) |
94
+ | solar_4_seconds | Energy | 1 | 4S | 60 | [Godahewa et al. 2021](https://openreview.net/pdf?id=wEc1mgAjU- ) |
95
+ | wind_4_seconds | Energy | 1 | 4S | 60 | [Godahewa et al. 2021](https://openreview.net/pdf?id=wEc1mgAjU- ) |
96
+ | rideshare | Transport | 2304 | 1H | 48 | [Godahewa et al. 2021](https://openreview.net/pdf?id=wEc1mgAjU- ) |
97
+ | oikolab_weather | Nature | 8 | 1H | 48 | [Oikolab](https://oikolab.com/) |
98
+ | temperature_rain | Nature | 32072 | 1D | 30 | [Godahewa et al. 2021](https://openreview.net/pdf?id=wEc1mgAjU- )
99
+
100
+
101
+ ### Dataset Usage
102
+
103
+ To load a particular dataset just specify its name from the table above e.g.:
104
+
105
+ ```python
106
+ load_dataset("monash_tsf", "nn5_daily")
107
+ ```
108
+ > Notes:
109
+ > - Data might contain missing values as in the original datasets.
110
+ > - The prediction length is either specified in the dataset or a default value depending on the frequency is used as in the original repository benchmark.
111
+
112
+
113
+ ### Supported Tasks and Leaderboards
114
+
115
+ #### `time-series-forecasting`
116
+
117
+ ##### `univariate-time-series-forecasting`
118
+
119
+ The univariate time series forecasting tasks involves learning the future one dimensional `target` values of a time series in a dataset for some `prediction_length` time steps. The performance of the forecast models can then be validated via the ground truth in the `validation` split and tested via the `test` split.
120
+
121
+ ##### `multivariate-time-series-forecasting`
122
+
123
+ The multivariate time series forecasting task involves learning the future vector of `target` values of a time series in a dataset for some `prediction_length` time steps. Similar to the univariate setting the performance of a multivariate model can be validated via the ground truth in the `validation` split and tested via the `test` split.
124
+
125
+ ### Languages
126
+
127
+ ## Dataset Structure
128
+
129
+ ### Data Instances
130
+
131
+ A sample from the training set is provided below:
132
+
133
+ ```python
134
+ {
135
+ 'start': datetime.datetime(2012, 1, 1, 0, 0),
136
+ 'target': [14.0, 18.0, 21.0, 20.0, 22.0, 20.0, ...],
137
+ 'feat_static_cat': [0],
138
+ 'feat_dynamic_real': [[0.3, 0.4], [0.1, 0.6], ...],
139
+ 'item_id': '0'
140
+ }
141
+ ```
142
+
143
+ ### Data Fields
144
+
145
+ For the univariate regular time series each series has the following keys:
146
+
147
+ * `start`: a datetime of the first entry of each time series in the dataset
148
+ * `target`: an array[float32] of the actual target values
149
+ * `feat_static_cat`: an array[uint64] which contains a categorical identifier of each time series in the dataset
150
+ * `feat_dynamic_real`: optional array of covariate features
151
+ * `item_id`: a string identifier of each time series in a dataset for reference
152
+
153
+ For the multivariate time series the `target` is a vector of the multivariate dimension for each time point.
154
+
155
+ ### Data Splits
156
+
157
+ The datasets are split in time depending on the prediction length specified in the datasets. In particular for each time series in a dataset there is a prediction length window of the future in the validation split and another prediction length more in the test split.
158
+
159
+
160
+ ## Dataset Creation
161
+
162
+ ### Curation Rationale
163
+
164
+ To facilitate the evaluation of global forecasting models. All datasets in our repository are intended for research purposes and to evaluate the performance of new forecasting algorithms.
165
+
166
+ ### Source Data
167
+
168
+ #### Initial Data Collection and Normalization
169
+
170
+ Out of the 30 datasets, 23 were already publicly available in different platforms with different data formats. The original sources of all datasets are mentioned in the datasets table above.
171
+
172
+ After extracting and curating these datasets, we analysed them individually to identify the datasets containing series with different frequencies and missing observations. Nine datasets contain time series belonging to different frequencies and the archive contains a separate dataset per each frequency.
173
+
174
+ #### Who are the source language producers?
175
+
176
+ The data comes from the datasets listed in the table above.
177
+
178
+ ### Annotations
179
+
180
+ #### Annotation process
181
+
182
+ The annotations come from the datasets listed in the table above.
183
+
184
+ #### Who are the annotators?
185
+
186
+ [More Information Needed]
187
+
188
+ ### Personal and Sensitive Information
189
+
190
+ [More Information Needed]
191
+
192
+ ## Considerations for Using the Data
193
+
194
+ ### Social Impact of Dataset
195
+
196
+ [More Information Needed]
197
+
198
+ ### Discussion of Biases
199
+
200
+ [More Information Needed]
201
+
202
+ ### Other Known Limitations
203
+
204
+ [More Information Needed]
205
+
206
+ ## Additional Information
207
+
208
+ ### Dataset Curators
209
+
210
+ * [Rakshitha Godahewa](mailto:rakshitha.godahewa@monash.edu)
211
+ * [Christoph Bergmeir](mailto:christoph.bergmeir@monash.edu)
212
+ * [Geoff Webb](mailto:geoff.webb@monash.edu)
213
+ * [Rob Hyndman](mailto:rob.hyndman@monash.edu)
214
+ * [Pablo Montero-Manso](mailto:pablo.monteromanso@sydney.edu.au)
215
+
216
+ ### Licensing Information
217
+
218
+ [Creative Commons Attribution 4.0 International](https://creativecommons.org/licenses/by/4.0/legalcode)
219
+
220
+ ### Citation Information
221
+
222
+ ```tex
223
+ @InProceedings{godahewa2021monash,
224
+ author = "Godahewa, Rakshitha and Bergmeir, Christoph and Webb, Geoffrey I. and Hyndman, Rob J. and Montero-Manso, Pablo",
225
+ title = "Monash Time Series Forecasting Archive",
226
+ booktitle = "Neural Information Processing Systems Track on Datasets and Benchmarks",
227
+ year = "2021",
228
+ note = "forthcoming"
229
+ }
230
+ ```
231
+
232
+ ### Contributions
233
+
234
+ Thanks to [@kashif](https://github.com/kashif) for adding this dataset.
dataset_infos.json ADDED
@@ -0,0 +1 @@
 
1
+ {"weather": {"description": "The first repository containing datasets of related time series for global forecasting.\n", "citation": "@InProceedings{godahewa2021monash,\n author = \"Godahewa, Rakshitha and Bergmeir, Christoph and Webb, Geoffrey I. and Hyndman, Rob J. and Montero-Manso, Pablo\",\n title = \"Monash Time Series Forecasting Archive\",\n booktitle = \"Neural Information Processing Systems Track on Datasets and Benchmarks\",\n year = \"2021\",\n note = \"forthcoming\"\n}\n", "homepage": "https://forecastingdata.org/", "license": "The Creative Commons Attribution 4.0 International License. https://creativecommons.org/licenses/by/4.0/", "features": {"start": {"dtype": "timestamp[s]", "id": null, "_type": "Value"}, "target": {"feature": {"dtype": "float32", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "feat_static_cat": {"feature": {"dtype": "uint64", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "feat_dynamic_real": {"feature": {"feature": {"dtype": "float32", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "length": -1, "id": null, "_type": "Sequence"}, "item_id": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "monash_tsf", "config_name": "weather", "version": {"version_str": "1.0.0", "description": null, "major": 1, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 176893738, "num_examples": 3010, "dataset_name": "monash_tsf"}, "test": {"name": "test", "num_bytes": 177638713, "num_examples": 3010, "dataset_name": "monash_tsf"}, "validation": {"name": "validation", "num_bytes": 177266226, "num_examples": 3010, "dataset_name": "monash_tsf"}}, "download_checksums": {"https://zenodo.org/record/4654822/files/weather_dataset.zip": {"num_bytes": 38820451, "checksum": "adc9195b4b842673892fec1ba22cced9c42d5c4cba9a261f455ffbe96db7847b"}}, "download_size": 38820451, "post_processing_size": null, "dataset_size": 531798677, "size_in_bytes": 570619128}, "tourism_yearly": {"description": "The first repository containing datasets of related time series for global forecasting.\n", "citation": "@InProceedings{godahewa2021monash,\n author = \"Godahewa, Rakshitha and Bergmeir, Christoph and Webb, Geoffrey I. and Hyndman, Rob J. and Montero-Manso, Pablo\",\n title = \"Monash Time Series Forecasting Archive\",\n booktitle = \"Neural Information Processing Systems Track on Datasets and Benchmarks\",\n year = \"2021\",\n note = \"forthcoming\"\n}\n", "homepage": "https://forecastingdata.org/", "license": "The Creative Commons Attribution 4.0 International License. https://creativecommons.org/licenses/by/4.0/", "features": {"start": {"dtype": "timestamp[s]", "id": null, "_type": "Value"}, "target": {"feature": {"dtype": "float32", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "feat_static_cat": {"feature": {"dtype": "uint64", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "feat_dynamic_real": {"feature": {"feature": {"dtype": "float32", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "length": -1, "id": null, "_type": "Sequence"}, "item_id": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "monash_tsf", "config_name": "tourism_yearly", "version": {"version_str": "1.0.0", "description": null, "major": 1, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 54264, "num_examples": 518, "dataset_name": "monash_tsf"}, "test": {"name": "test", "num_bytes": 71358, "num_examples": 518, "dataset_name": "monash_tsf"}, "validation": {"name": "validation", "num_bytes": 62811, "num_examples": 518, "dataset_name": "monash_tsf"}}, "download_checksums": {"https://zenodo.org/record/4656103/files/tourism_yearly_dataset.zip": {"num_bytes": 36749, "checksum": "20238ddb9b11465b061bbfadcc398ef01f0896a8791199381b3edc706abf0263"}}, "download_size": 36749, "post_processing_size": null, "dataset_size": 188433, "size_in_bytes": 225182}, "tourism_quarterly": {"description": "The first repository containing datasets of related time series for global forecasting.\n", "citation": "@InProceedings{godahewa2021monash,\n author = \"Godahewa, Rakshitha and Bergmeir, Christoph and Webb, Geoffrey I. and Hyndman, Rob J. and Montero-Manso, Pablo\",\n title = \"Monash Time Series Forecasting Archive\",\n booktitle = \"Neural Information Processing Systems Track on Datasets and Benchmarks\",\n year = \"2021\",\n note = \"forthcoming\"\n}\n", "homepage": "https://forecastingdata.org/", "license": "The Creative Commons Attribution 4.0 International License. https://creativecommons.org/licenses/by/4.0/", "features": {"start": {"dtype": "timestamp[s]", "id": null, "_type": "Value"}, "target": {"feature": {"dtype": "float32", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "feat_static_cat": {"feature": {"dtype": "uint64", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "feat_dynamic_real": {"feature": {"feature": {"dtype": "float32", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "length": -1, "id": null, "_type": "Sequence"}, "item_id": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "monash_tsf", "config_name": "tourism_quarterly", "version": {"version_str": "1.0.0", "description": null, "major": 1, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 162738, "num_examples": 427, "dataset_name": "monash_tsf"}, "test": {"name": "test", "num_bytes": 190920, "num_examples": 427, "dataset_name": "monash_tsf"}, "validation": {"name": "validation", "num_bytes": 176829, "num_examples": 427, "dataset_name": "monash_tsf"}}, "download_checksums": {"https://zenodo.org/record/4656093/files/tourism_quarterly_dataset.zip": {"num_bytes": 93833, "checksum": "fa16da64f29aa04df447756cb65a5d58f8bc3d860fc22751c76850a3ffb048c3"}}, "download_size": 93833, "post_processing_size": null, "dataset_size": 530487, "size_in_bytes": 624320}, "tourism_monthly": {"description": "The first repository containing datasets of related time series for global forecasting.\n", "citation": "@InProceedings{godahewa2021monash,\n author = \"Godahewa, Rakshitha and Bergmeir, Christoph and Webb, Geoffrey I. and Hyndman, Rob J. and Montero-Manso, Pablo\",\n title = \"Monash Time Series Forecasting Archive\",\n booktitle = \"Neural Information Processing Systems Track on Datasets and Benchmarks\",\n year = \"2021\",\n note = \"forthcoming\"\n}\n", "homepage": "https://forecastingdata.org/", "license": "The Creative Commons Attribution 4.0 International License. https://creativecommons.org/licenses/by/4.0/", "features": {"start": {"dtype": "timestamp[s]", "id": null, "_type": "Value"}, "target": {"feature": {"dtype": "float32", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "feat_static_cat": {"feature": {"dtype": "uint64", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "feat_dynamic_real": {"feature": {"feature": {"dtype": "float32", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "length": -1, "id": null, "_type": "Sequence"}, "item_id": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "monash_tsf", "config_name": "tourism_monthly", "version": {"version_str": "1.0.0", "description": null, "major": 1, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 391518, "num_examples": 366, "dataset_name": "monash_tsf"}, "test": {"name": "test", "num_bytes": 463986, "num_examples": 366, "dataset_name": "monash_tsf"}, "validation": {"name": "validation", "num_bytes": 427752, "num_examples": 366, "dataset_name": "monash_tsf"}}, "download_checksums": {"https://zenodo.org/record/4656096/files/tourism_monthly_dataset.zip": {"num_bytes": 199791, "checksum": "6e434ad8a0ef6acb55ebee0e1d1ffe22707ea4e62c05e091326267f3a5b4a93c"}}, "download_size": 199791, "post_processing_size": null, "dataset_size": 1283256, "size_in_bytes": 1483047}, "cif_2016": {"description": "The first repository containing datasets of related time series for global forecasting.\n", "citation": "@InProceedings{godahewa2021monash,\n author = \"Godahewa, Rakshitha and Bergmeir, Christoph and Webb, Geoffrey I. and Hyndman, Rob J. and Montero-Manso, Pablo\",\n title = \"Monash Time Series Forecasting Archive\",\n booktitle = \"Neural Information Processing Systems Track on Datasets and Benchmarks\",\n year = \"2021\",\n note = \"forthcoming\"\n}\n", "homepage": "https://forecastingdata.org/", "license": "The Creative Commons Attribution 4.0 International License. https://creativecommons.org/licenses/by/4.0/", "features": {"start": {"dtype": "timestamp[s]", "id": null, "_type": "Value"}, "target": {"feature": {"dtype": "float32", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "feat_static_cat": {"feature": {"dtype": "uint64", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "feat_dynamic_real": {"feature": {"feature": {"dtype": "float32", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "length": -1, "id": null, "_type": "Sequence"}, "item_id": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "monash_tsf", "config_name": "cif_2016", "version": {"version_str": "1.0.0", "description": null, "major": 1, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 24731, "num_examples": 72, "dataset_name": "monash_tsf"}, "test": {"name": "test", "num_bytes": 31859, "num_examples": 72, "dataset_name": "monash_tsf"}, "validation": {"name": "validation", "num_bytes": 28295, "num_examples": 72, "dataset_name": "monash_tsf"}}, "download_checksums": {"https://zenodo.org/record/4656042/files/cif_2016_dataset.zip": {"num_bytes": 53344, "checksum": "e0a7d93d6fb66dd2bb3ceb23bec73c2f39fe9cec909b4599ca35fef22b8d8f59"}}, "download_size": 53344, "post_processing_size": null, "dataset_size": 84885, "size_in_bytes": 138229}, "london_smart_meters": {"description": "The first repository containing datasets of related time series for global forecasting.\n", "citation": "@InProceedings{godahewa2021monash,\n author = \"Godahewa, Rakshitha and Bergmeir, Christoph and Webb, Geoffrey I. and Hyndman, Rob J. and Montero-Manso, Pablo\",\n title = \"Monash Time Series Forecasting Archive\",\n booktitle = \"Neural Information Processing Systems Track on Datasets and Benchmarks\",\n year = \"2021\",\n note = \"forthcoming\"\n}\n", "homepage": "https://forecastingdata.org/", "license": "The Creative Commons Attribution 4.0 International License. https://creativecommons.org/licenses/by/4.0/", "features": {"start": {"dtype": "timestamp[s]", "id": null, "_type": "Value"}, "target": {"feature": {"dtype": "float32", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "feat_static_cat": {"feature": {"dtype": "uint64", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "feat_dynamic_real": {"feature": {"feature": {"dtype": "float32", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "length": -1, "id": null, "_type": "Sequence"}, "item_id": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "monash_tsf", "config_name": "london_smart_meters", "version": {"version_str": "1.0.0", "description": null, "major": 1, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 684386194, "num_examples": 5560, "dataset_name": "monash_tsf"}, "test": {"name": "test", "num_bytes": 687138394, "num_examples": 5560, "dataset_name": "monash_tsf"}, "validation": {"name": "validation", "num_bytes": 685762294, "num_examples": 5560, "dataset_name": "monash_tsf"}}, "download_checksums": {"https://zenodo.org/record/4656072/files/london_smart_meters_dataset_with_missing_values.zip": {"num_bytes": 219673439, "checksum": "20a0ead3ad2ed7c073a9372062cbd473fdbbcf5d86d73d1c97bd3909a0cf7ac8"}}, "download_size": 219673439, "post_processing_size": null, "dataset_size": 2057286882, "size_in_bytes": 2276960321}, "australian_electricity_demand": {"description": "The first repository containing datasets of related time series for global forecasting.\n", "citation": "@InProceedings{godahewa2021monash,\n author = \"Godahewa, Rakshitha and Bergmeir, Christoph and Webb, Geoffrey I. and Hyndman, Rob J. and Montero-Manso, Pablo\",\n title = \"Monash Time Series Forecasting Archive\",\n booktitle = \"Neural Information Processing Systems Track on Datasets and Benchmarks\",\n year = \"2021\",\n note = \"forthcoming\"\n}\n", "homepage": "https://forecastingdata.org/", "license": "The Creative Commons Attribution 4.0 International License. https://creativecommons.org/licenses/by/4.0/", "features": {"start": {"dtype": "timestamp[s]", "id": null, "_type": "Value"}, "target": {"feature": {"dtype": "float32", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "feat_static_cat": {"feature": {"dtype": "uint64", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "feat_dynamic_real": {"feature": {"feature": {"dtype": "float32", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "length": -1, "id": null, "_type": "Sequence"}, "item_id": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "monash_tsf", "config_name": "australian_electricity_demand", "version": {"version_str": "1.0.0", "description": null, "major": 1, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 4763162, "num_examples": 5, "dataset_name": "monash_tsf"}, "test": {"name": "test", "num_bytes": 4765637, "num_examples": 5, "dataset_name": "monash_tsf"}, "validation": {"name": "validation", "num_bytes": 4764400, "num_examples": 5, "dataset_name": "monash_tsf"}}, "download_checksums": {"https://zenodo.org/record/4659727/files/australian_electricity_demand_dataset.zip": {"num_bytes": 5770526, "checksum": "b1439c28a631766bd05ee327f1f430a887b9f53b38a8cd5c0769ded1fd4aaf5a"}}, "download_size": 5770526, "post_processing_size": null, "dataset_size": 14293199, "size_in_bytes": 20063725}, "wind_farms_minutely": {"description": "The first repository containing datasets of related time series for global forecasting.\n", "citation": "@InProceedings{godahewa2021monash,\n author = \"Godahewa, Rakshitha and Bergmeir, Christoph and Webb, Geoffrey I. and Hyndman, Rob J. and Montero-Manso, Pablo\",\n title = \"Monash Time Series Forecasting Archive\",\n booktitle = \"Neural Information Processing Systems Track on Datasets and Benchmarks\",\n year = \"2021\",\n note = \"forthcoming\"\n}\n", "homepage": "https://forecastingdata.org/", "license": "The Creative Commons Attribution 4.0 International License. https://creativecommons.org/licenses/by/4.0/", "features": {"start": {"dtype": "timestamp[s]", "id": null, "_type": "Value"}, "target": {"feature": {"dtype": "float32", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "feat_static_cat": {"feature": {"dtype": "uint64", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "feat_dynamic_real": {"feature": {"feature": {"dtype": "float32", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "length": -1, "id": null, "_type": "Sequence"}, "item_id": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "monash_tsf", "config_name": "wind_farms_minutely", "version": {"version_str": "1.0.0", "description": null, "major": 1, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 710078918, "num_examples": 339, "dataset_name": "monash_tsf"}, "test": {"name": "test", "num_bytes": 710246723, "num_examples": 339, "dataset_name": "monash_tsf"}, "validation": {"name": "validation", "num_bytes": 710162820, "num_examples": 339, "dataset_name": "monash_tsf"}}, "download_checksums": {"https://zenodo.org/record/4654909/files/wind_farms_minutely_dataset_with_missing_values.zip": {"num_bytes": 71383130, "checksum": "4633a7255b6720347fc682003b5441e7bbd1a1097a4834cac0492852fb42e47a"}}, "download_size": 71383130, "post_processing_size": null, "dataset_size": 2130488461, "size_in_bytes": 2201871591}, "bitcoin": {"description": "The first repository containing datasets of related time series for global forecasting.\n", "citation": "@InProceedings{godahewa2021monash,\n author = \"Godahewa, Rakshitha and Bergmeir, Christoph and Webb, Geoffrey I. and Hyndman, Rob J. and Montero-Manso, Pablo\",\n title = \"Monash Time Series Forecasting Archive\",\n booktitle = \"Neural Information Processing Systems Track on Datasets and Benchmarks\",\n year = \"2021\",\n note = \"forthcoming\"\n}\n", "homepage": "https://forecastingdata.org/", "license": "The Creative Commons Attribution 4.0 International License. https://creativecommons.org/licenses/by/4.0/", "features": {"start": {"dtype": "timestamp[s]", "id": null, "_type": "Value"}, "target": {"feature": {"dtype": "float32", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "feat_static_cat": {"feature": {"dtype": "uint64", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "feat_dynamic_real": {"feature": {"feature": {"dtype": "float32", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "length": -1, "id": null, "_type": "Sequence"}, "item_id": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "monash_tsf", "config_name": "bitcoin", "version": {"version_str": "1.0.0", "description": null, "major": 1, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 336511, "num_examples": 18, "dataset_name": "monash_tsf"}, "test": {"name": "test", "num_bytes": 340966, "num_examples": 18, "dataset_name": "monash_tsf"}, "validation": {"name": "validation", "num_bytes": 338738, "num_examples": 18, "dataset_name": "monash_tsf"}}, "download_checksums": {"https://zenodo.org/record/5121965/files/bitcoin_dataset_with_missing_values.zip": {"num_bytes": 220403, "checksum": "aca43bae943dbf24617e885b5165f2493578fb195a888702ec85947e727f015f"}}, "download_size": 220403, "post_processing_size": null, "dataset_size": 1016215, "size_in_bytes": 1236618}, "pedestrian_counts": {"description": "The first repository containing datasets of related time series for global forecasting.\n", "citation": "@InProceedings{godahewa2021monash,\n author = \"Godahewa, Rakshitha and Bergmeir, Christoph and Webb, Geoffrey I. and Hyndman, Rob J. and Montero-Manso, Pablo\",\n title = \"Monash Time Series Forecasting Archive\",\n booktitle = \"Neural Information Processing Systems Track on Datasets and Benchmarks\",\n year = \"2021\",\n note = \"forthcoming\"\n}\n", "homepage": "https://forecastingdata.org/", "license": "The Creative Commons Attribution 4.0 International License. https://creativecommons.org/licenses/by/4.0/", "features": {"start": {"dtype": "timestamp[s]", "id": null, "_type": "Value"}, "target": {"feature": {"dtype": "float32", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "feat_static_cat": {"feature": {"dtype": "uint64", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "feat_dynamic_real": {"feature": {"feature": {"dtype": "float32", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "length": -1, "id": null, "_type": "Sequence"}, "item_id": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "monash_tsf", "config_name": "pedestrian_counts", "version": {"version_str": "1.0.0", "description": null, "major": 1, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 12897120, "num_examples": 66, "dataset_name": "monash_tsf"}, "test": {"name": "test", "num_bytes": 12923256, "num_examples": 66, "dataset_name": "monash_tsf"}, "validation": {"name": "validation", "num_bytes": 12910188, "num_examples": 66, "dataset_name": "monash_tsf"}}, "download_checksums": {"https://zenodo.org/record/4656626/files/pedestrian_counts_dataset.zip": {"num_bytes": 4587054, "checksum": "6e81cb8cad43650e7e0f754a1103fc75c960e03387bae9b91146aa3241c9aa50"}}, "download_size": 4587054, "post_processing_size": null, "dataset_size": 38730564, "size_in_bytes": 43317618}, "vehicle_trips": {"description": "The first repository containing datasets of related time series for global forecasting.\n", "citation": "@InProceedings{godahewa2021monash,\n author = \"Godahewa, Rakshitha and Bergmeir, Christoph and Webb, Geoffrey I. and Hyndman, Rob J. and Montero-Manso, Pablo\",\n title = \"Monash Time Series Forecasting Archive\",\n booktitle = \"Neural Information Processing Systems Track on Datasets and Benchmarks\",\n year = \"2021\",\n note = \"forthcoming\"\n}\n", "homepage": "https://forecastingdata.org/", "license": "The Creative Commons Attribution 4.0 International License. https://creativecommons.org/licenses/by/4.0/", "features": {"start": {"dtype": "timestamp[s]", "id": null, "_type": "Value"}, "target": {"feature": {"dtype": "float32", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "feat_static_cat": {"feature": {"dtype": "uint64", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "feat_dynamic_real": {"feature": {"feature": {"dtype": "float32", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "length": -1, "id": null, "_type": "Sequence"}, "item_id": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "monash_tsf", "config_name": "vehicle_trips", "version": {"version_str": "1.0.0", "description": null, "major": 1, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 105261, "num_examples": 329, "dataset_name": "monash_tsf"}, "test": {"name": "test", "num_bytes": 186688, "num_examples": 329, "dataset_name": "monash_tsf"}, "validation": {"name": "validation", "num_bytes": 145974, "num_examples": 329, "dataset_name": "monash_tsf"}}, "download_checksums": {"https://zenodo.org/record/5122535/files/vehicle_trips_dataset_with_missing_values.zip": {"num_bytes": 44914, "checksum": "f0e4df4d85705567cfe5489267f3fdbb398155e1d29796d2b4690cbbfd054730"}}, "download_size": 44914, "post_processing_size": null, "dataset_size": 437923, "size_in_bytes": 482837}, "kdd_cup_2018": {"description": "The first repository containing datasets of related time series for global forecasting.\n", "citation": "@InProceedings{godahewa2021monash,\n author = \"Godahewa, Rakshitha and Bergmeir, Christoph and Webb, Geoffrey I. and Hyndman, Rob J. and Montero-Manso, Pablo\",\n title = \"Monash Time Series Forecasting Archive\",\n booktitle = \"Neural Information Processing Systems Track on Datasets and Benchmarks\",\n year = \"2021\",\n note = \"forthcoming\"\n}\n", "homepage": "https://forecastingdata.org/", "license": "The Creative Commons Attribution 4.0 International License. https://creativecommons.org/licenses/by/4.0/", "features": {"start": {"dtype": "timestamp[s]", "id": null, "_type": "Value"}, "target": {"feature": {"dtype": "float32", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "feat_static_cat": {"feature": {"dtype": "uint64", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "feat_dynamic_real": {"feature": {"feature": {"dtype": "float32", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "length": -1, "id": null, "_type": "Sequence"}, "item_id": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "monash_tsf", "config_name": "kdd_cup_2018", "version": {"version_str": "1.0.0", "description": null, "major": 1, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 12040046, "num_examples": 270, "dataset_name": "monash_tsf"}, "test": {"name": "test", "num_bytes": 12146966, "num_examples": 270, "dataset_name": "monash_tsf"}, "validation": {"name": "validation", "num_bytes": 12093506, "num_examples": 270, "dataset_name": "monash_tsf"}}, "download_checksums": {"https://zenodo.org/record/4656719/files/kdd_cup_2018_dataset_with_missing_values.zip": {"num_bytes": 2456948, "checksum": "3d0d2dbec3621b20c7a8ae6e6f00a63bfaebb962dca34d9f60f0ddbbde03e9bd"}}, "download_size": 2456948, "post_processing_size": null, "dataset_size": 36280518, "size_in_bytes": 38737466}, "nn5_daily": {"description": "The first repository containing datasets of related time series for global forecasting.\n", "citation": "@InProceedings{godahewa2021monash,\n author = \"Godahewa, Rakshitha and Bergmeir, Christoph and Webb, Geoffrey I. and Hyndman, Rob J. and Montero-Manso, Pablo\",\n title = \"Monash Time Series Forecasting Archive\",\n booktitle = \"Neural Information Processing Systems Track on Datasets and Benchmarks\",\n year = \"2021\",\n note = \"forthcoming\"\n}\n", "homepage": "https://forecastingdata.org/", "license": "The Creative Commons Attribution 4.0 International License. https://creativecommons.org/licenses/by/4.0/", "features": {"start": {"dtype": "timestamp[s]", "id": null, "_type": "Value"}, "target": {"feature": {"dtype": "float32", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "feat_static_cat": {"feature": {"dtype": "uint64", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "feat_dynamic_real": {"feature": {"feature": {"dtype": "float32", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "length": -1, "id": null, "_type": "Sequence"}, "item_id": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "monash_tsf", "config_name": "nn5_daily", "version": {"version_str": "1.0.0", "description": null, "major": 1, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 314828, "num_examples": 111, "dataset_name": "monash_tsf"}, "test": {"name": "test", "num_bytes": 366110, "num_examples": 111, "dataset_name": "monash_tsf"}, "validation": {"name": "validation", "num_bytes": 340469, "num_examples": 111, "dataset_name": "monash_tsf"}}, "download_checksums": {"https://zenodo.org/record/4656110/files/nn5_daily_dataset_with_missing_values.zip": {"num_bytes": 287708, "checksum": "7022d158f6e49e323f3bd76f48fb18ba78b6d013972a97980af01c56f6f4c74b"}}, "download_size": 287708, "post_processing_size": null, "dataset_size": 1021407, "size_in_bytes": 1309115}, "nn5_weekly": {"description": "The first repository containing datasets of related time series for global forecasting.\n", "citation": "@InProceedings{godahewa2021monash,\n author = \"Godahewa, Rakshitha and Bergmeir, Christoph and Webb, Geoffrey I. and Hyndman, Rob J. and Montero-Manso, Pablo\",\n title = \"Monash Time Series Forecasting Archive\",\n booktitle = \"Neural Information Processing Systems Track on Datasets and Benchmarks\",\n year = \"2021\",\n note = \"forthcoming\"\n}\n", "homepage": "https://forecastingdata.org/", "license": "The Creative Commons Attribution 4.0 International License. https://creativecommons.org/licenses/by/4.0/", "features": {"start": {"dtype": "timestamp[s]", "id": null, "_type": "Value"}, "target": {"feature": {"dtype": "float32", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "feat_static_cat": {"feature": {"dtype": "uint64", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "feat_dynamic_real": {"feature": {"feature": {"dtype": "float32", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "length": -1, "id": null, "_type": "Sequence"}, "item_id": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "monash_tsf", "config_name": "nn5_weekly", "version": {"version_str": "1.0.0", "description": null, "major": 1, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 48344, "num_examples": 111, "dataset_name": "monash_tsf"}, "test": {"name": "test", "num_bytes": 55670, "num_examples": 111, "dataset_name": "monash_tsf"}, "validation": {"name": "validation", "num_bytes": 52007, "num_examples": 111, "dataset_name": "monash_tsf"}}, "download_checksums": {"https://zenodo.org/record/4656125/files/nn5_weekly_dataset.zip": {"num_bytes": 62043, "checksum": "ce6ad3aee8cb6b665a587a40c7884e9e23d1b0f238a32dc2bdc34de7a57f3d54"}}, "download_size": 62043, "post_processing_size": null, "dataset_size": 156021, "size_in_bytes": 218064}, "kaggle_web_traffic": {"description": "The first repository containing datasets of related time series for global forecasting.\n", "citation": "@InProceedings{godahewa2021monash,\n author = \"Godahewa, Rakshitha and Bergmeir, Christoph and Webb, Geoffrey I. and Hyndman, Rob J. and Montero-Manso, Pablo\",\n title = \"Monash Time Series Forecasting Archive\",\n booktitle = \"Neural Information Processing Systems Track on Datasets and Benchmarks\",\n year = \"2021\",\n note = \"forthcoming\"\n}\n", "homepage": "https://forecastingdata.org/", "license": "The Creative Commons Attribution 4.0 International License. https://creativecommons.org/licenses/by/4.0/", "features": {"start": {"dtype": "timestamp[s]", "id": null, "_type": "Value"}, "target": {"feature": {"dtype": "float32", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "feat_static_cat": {"feature": {"dtype": "uint64", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "feat_dynamic_real": {"feature": {"feature": {"dtype": "float32", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "length": -1, "id": null, "_type": "Sequence"}, "item_id": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "monash_tsf", "config_name": "kaggle_web_traffic", "version": {"version_str": "1.0.0", "description": null, "major": 1, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 415494391, "num_examples": 145063, "dataset_name": "monash_tsf"}, "test": {"name": "test", "num_bytes": 486103806, "num_examples": 145063, "dataset_name": "monash_tsf"}, "validation": {"name": "validation", "num_bytes": 450799098, "num_examples": 145063, "dataset_name": "monash_tsf"}}, "download_checksums": {"https://zenodo.org/record/4656080/files/kaggle_web_traffic_dataset_with_missing_values.zip": {"num_bytes": 145485324, "checksum": "44e10fc187475f673e2aac837f5f54cf32333fed25222d006f287355ec8699a9"}}, "download_size": 145485324, "post_processing_size": null, "dataset_size": 1352397295, "size_in_bytes": 1497882619}, "kaggle_web_traffic_weekly": {"description": "The first repository containing datasets of related time series for global forecasting.\n", "citation": "@InProceedings{godahewa2021monash,\n author = \"Godahewa, Rakshitha and Bergmeir, Christoph and Webb, Geoffrey I. and Hyndman, Rob J. and Montero-Manso, Pablo\",\n title = \"Monash Time Series Forecasting Archive\",\n booktitle = \"Neural Information Processing Systems Track on Datasets and Benchmarks\",\n year = \"2021\",\n note = \"forthcoming\"\n}\n", "homepage": "https://forecastingdata.org/", "license": "The Creative Commons Attribution 4.0 International License. https://creativecommons.org/licenses/by/4.0/", "features": {"start": {"dtype": "timestamp[s]", "id": null, "_type": "Value"}, "target": {"feature": {"dtype": "float32", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "feat_static_cat": {"feature": {"dtype": "uint64", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "feat_dynamic_real": {"feature": {"feature": {"dtype": "float32", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "length": -1, "id": null, "_type": "Sequence"}, "item_id": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "monash_tsf", "config_name": "kaggle_web_traffic_weekly", "version": {"version_str": "1.0.0", "description": null, "major": 1, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 64242469, "num_examples": 145063, "dataset_name": "monash_tsf"}, "test": {"name": "test", "num_bytes": 73816627, "num_examples": 145063, "dataset_name": "monash_tsf"}, "validation": {"name": "validation", "num_bytes": 69029548, "num_examples": 145063, "dataset_name": "monash_tsf"}}, "download_checksums": {"https://zenodo.org/record/4656664/files/kaggle_web_traffic_weekly_dataset.zip": {"num_bytes": 28930900, "checksum": "5d286ff0c9e4485eae74525edec9540ff4b689ec9b580cf9e85c8ab7225d29c1"}}, "download_size": 28930900, "post_processing_size": null, "dataset_size": 207088644, "size_in_bytes": 236019544}, "solar_10_minutes": {"description": "The first repository containing datasets of related time series for global forecasting.\n", "citation": "@InProceedings{godahewa2021monash,\n author = \"Godahewa, Rakshitha and Bergmeir, Christoph and Webb, Geoffrey I. and Hyndman, Rob J. and Montero-Manso, Pablo\",\n title = \"Monash Time Series Forecasting Archive\",\n booktitle = \"Neural Information Processing Systems Track on Datasets and Benchmarks\",\n year = \"2021\",\n note = \"forthcoming\"\n}\n", "homepage": "https://forecastingdata.org/", "license": "The Creative Commons Attribution 4.0 International License. https://creativecommons.org/licenses/by/4.0/", "features": {"start": {"dtype": "timestamp[s]", "id": null, "_type": "Value"}, "target": {"feature": {"dtype": "float32", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "feat_static_cat": {"feature": {"dtype": "uint64", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "feat_dynamic_real": {"feature": {"feature": {"dtype": "float32", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "length": -1, "id": null, "_type": "Sequence"}, "item_id": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "monash_tsf", "config_name": "solar_10_minutes", "version": {"version_str": "1.0.0", "description": null, "major": 1, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 29640033, "num_examples": 137, "dataset_name": "monash_tsf"}, "test": {"name": "test", "num_bytes": 29707848, "num_examples": 137, "dataset_name": "monash_tsf"}, "validation": {"name": "validation", "num_bytes": 29673941, "num_examples": 137, "dataset_name": "monash_tsf"}}, "download_checksums": {"https://zenodo.org/record/4656144/files/solar_10_minutes_dataset.zip": {"num_bytes": 4559353, "checksum": "63ec43cefc45d87af17f06255daad8343cc2593d3572e4d7178d791ab9387390"}}, "download_size": 4559353, "post_processing_size": null, "dataset_size": 89021822, "size_in_bytes": 93581175}, "solar_weekly": {"description": "The first repository containing datasets of related time series for global forecasting.\n", "citation": "@InProceedings{godahewa2021monash,\n author = \"Godahewa, Rakshitha and Bergmeir, Christoph and Webb, Geoffrey I. and Hyndman, Rob J. and Montero-Manso, Pablo\",\n title = \"Monash Time Series Forecasting Archive\",\n booktitle = \"Neural Information Processing Systems Track on Datasets and Benchmarks\",\n year = \"2021\",\n note = \"forthcoming\"\n}\n", "homepage": "https://forecastingdata.org/", "license": "The Creative Commons Attribution 4.0 International License. https://creativecommons.org/licenses/by/4.0/", "features": {"start": {"dtype": "timestamp[s]", "id": null, "_type": "Value"}, "target": {"feature": {"dtype": "float32", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "feat_static_cat": {"feature": {"dtype": "uint64", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "feat_dynamic_real": {"feature": {"feature": {"dtype": "float32", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "length": -1, "id": null, "_type": "Sequence"}, "item_id": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "monash_tsf", "config_name": "solar_weekly", "version": {"version_str": "1.0.0", "description": null, "major": 1, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 28614, "num_examples": 137, "dataset_name": "monash_tsf"}, "test": {"name": "test", "num_bytes": 34265, "num_examples": 137, "dataset_name": "monash_tsf"}, "validation": {"name": "validation", "num_bytes": 31439, "num_examples": 137, "dataset_name": "monash_tsf"}}, "download_checksums": {"https://zenodo.org/record/4656151/files/solar_weekly_dataset.zip": {"num_bytes": 24375, "checksum": "421426b22c22997697aeeaadb8e71cc73a7a0a36ed8097037ff6b378650f428f"}}, "download_size": 24375, "post_processing_size": null, "dataset_size": 94318, "size_in_bytes": 118693}, "car_parts": {"description": "The first repository containing datasets of related time series for global forecasting.\n", "citation": "@InProceedings{godahewa2021monash,\n author = \"Godahewa, Rakshitha and Bergmeir, Christoph and Webb, Geoffrey I. and Hyndman, Rob J. and Montero-Manso, Pablo\",\n title = \"Monash Time Series Forecasting Archive\",\n booktitle = \"Neural Information Processing Systems Track on Datasets and Benchmarks\",\n year = \"2021\",\n note = \"forthcoming\"\n}\n", "homepage": "https://forecastingdata.org/", "license": "The Creative Commons Attribution 4.0 International License. https://creativecommons.org/licenses/by/4.0/", "features": {"start": {"dtype": "timestamp[s]", "id": null, "_type": "Value"}, "target": {"feature": {"dtype": "float32", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "feat_static_cat": {"feature": {"dtype": "uint64", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "feat_dynamic_real": {"feature": {"feature": {"dtype": "float32", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "length": -1, "id": null, "_type": "Sequence"}, "item_id": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "monash_tsf", "config_name": "car_parts", "version": {"version_str": "1.0.0", "description": null, "major": 1, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 396653, "num_examples": 2674, "dataset_name": "monash_tsf"}, "test": {"name": "test", "num_bytes": 661379, "num_examples": 2674, "dataset_name": "monash_tsf"}, "validation": {"name": "validation", "num_bytes": 529016, "num_examples": 2674, "dataset_name": "monash_tsf"}}, "download_checksums": {"https://zenodo.org/record/4656022/files/car_parts_dataset_with_missing_values.zip": {"num_bytes": 39656, "checksum": "80557703397c7dca6c007bd28930b0e213e529f428afd02e5af87804fdc38ee1"}}, "download_size": 39656, "post_processing_size": null, "dataset_size": 1587048, "size_in_bytes": 1626704}, "fred_md": {"description": "The first repository containing datasets of related time series for global forecasting.\n", "citation": "@InProceedings{godahewa2021monash,\n author = \"Godahewa, Rakshitha and Bergmeir, Christoph and Webb, Geoffrey I. and Hyndman, Rob J. and Montero-Manso, Pablo\",\n title = \"Monash Time Series Forecasting Archive\",\n booktitle = \"Neural Information Processing Systems Track on Datasets and Benchmarks\",\n year = \"2021\",\n note = \"forthcoming\"\n}\n", "homepage": "https://forecastingdata.org/", "license": "The Creative Commons Attribution 4.0 International License. https://creativecommons.org/licenses/by/4.0/", "features": {"start": {"dtype": "timestamp[s]", "id": null, "_type": "Value"}, "target": {"feature": {"dtype": "float32", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "feat_static_cat": {"feature": {"dtype": "uint64", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "feat_dynamic_real": {"feature": {"feature": {"dtype": "float32", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "length": -1, "id": null, "_type": "Sequence"}, "item_id": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "monash_tsf", "config_name": "fred_md", "version": {"version_str": "1.0.0", "description": null, "major": 1, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 314514, "num_examples": 107, "dataset_name": "monash_tsf"}, "test": {"name": "test", "num_bytes": 325107, "num_examples": 107, "dataset_name": "monash_tsf"}, "validation": {"name": "validation", "num_bytes": 319811, "num_examples": 107, "dataset_name": "monash_tsf"}}, "download_checksums": {"https://zenodo.org/record/4654833/files/fred_md_dataset.zip": {"num_bytes": 169107, "checksum": "305c0edd2b5e97159c6339be4990ea97fdf86772c1edef2a0dfc836bf29f45c3"}}, "download_size": 169107, "post_processing_size": null, "dataset_size": 959432, "size_in_bytes": 1128539}, "traffic_hourly": {"description": "The first repository containing datasets of related time series for global forecasting.\n", "citation": "@InProceedings{godahewa2021monash,\n author = \"Godahewa, Rakshitha and Bergmeir, Christoph and Webb, Geoffrey I. and Hyndman, Rob J. and Montero-Manso, Pablo\",\n title = \"Monash Time Series Forecasting Archive\",\n booktitle = \"Neural Information Processing Systems Track on Datasets and Benchmarks\",\n year = \"2021\",\n note = \"forthcoming\"\n}\n", "homepage": "https://forecastingdata.org/", "license": "The Creative Commons Attribution 4.0 International License. https://creativecommons.org/licenses/by/4.0/", "features": {"start": {"dtype": "timestamp[s]", "id": null, "_type": "Value"}, "target": {"feature": {"dtype": "float32", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "feat_static_cat": {"feature": {"dtype": "uint64", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "feat_dynamic_real": {"feature": {"feature": {"dtype": "float32", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "length": -1, "id": null, "_type": "Sequence"}, "item_id": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "monash_tsf", "config_name": "traffic_hourly", "version": {"version_str": "1.0.0", "description": null, "major": 1, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 62071974, "num_examples": 862, "dataset_name": "monash_tsf"}, "test": {"name": "test", "num_bytes": 62413326, "num_examples": 862, "dataset_name": "monash_tsf"}, "validation": {"name": "validation", "num_bytes": 62242650, "num_examples": 862, "dataset_name": "monash_tsf"}}, "download_checksums": {"https://zenodo.org/record/4656132/files/traffic_hourly_dataset.zip": {"num_bytes": 22868806, "checksum": "3db12ba866a9c9d3c8109b7b6d189a990c38d0e5002fa2617022157358d08299"}}, "download_size": 22868806, "post_processing_size": null, "dataset_size": 186727950, "size_in_bytes": 209596756}, "traffic_weekly": {"description": "The first repository containing datasets of related time series for global forecasting.\n", "citation": "@InProceedings{godahewa2021monash,\n author = \"Godahewa, Rakshitha and Bergmeir, Christoph and Webb, Geoffrey I. and Hyndman, Rob J. and Montero-Manso, Pablo\",\n title = \"Monash Time Series Forecasting Archive\",\n booktitle = \"Neural Information Processing Systems Track on Datasets and Benchmarks\",\n year = \"2021\",\n note = \"forthcoming\"\n}\n", "homepage": "https://forecastingdata.org/", "license": "The Creative Commons Attribution 4.0 International License. https://creativecommons.org/licenses/by/4.0/", "features": {"start": {"dtype": "timestamp[s]", "id": null, "_type": "Value"}, "target": {"feature": {"dtype": "float32", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "feat_static_cat": {"feature": {"dtype": "uint64", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "feat_dynamic_real": {"feature": {"feature": {"dtype": "float32", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "length": -1, "id": null, "_type": "Sequence"}, "item_id": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "monash_tsf", "config_name": "traffic_weekly", "version": {"version_str": "1.0.0", "description": null, "major": 1, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 344154, "num_examples": 862, "dataset_name": "monash_tsf"}, "test": {"name": "test", "num_bytes": 401046, "num_examples": 862, "dataset_name": "monash_tsf"}, "validation": {"name": "validation", "num_bytes": 372600, "num_examples": 862, "dataset_name": "monash_tsf"}}, "download_checksums": {"https://zenodo.org/record/4656135/files/traffic_weekly_dataset.zip": {"num_bytes": 245126, "checksum": "df3fc458fb1d6f770451fc45c43b930c091e69e86d508410dae4843918851525"}}, "download_size": 245126, "post_processing_size": null, "dataset_size": 1117800, "size_in_bytes": 1362926}, "hospital": {"description": "The first repository containing datasets of related time series for global forecasting.\n", "citation": "@InProceedings{godahewa2021monash,\n author = \"Godahewa, Rakshitha and Bergmeir, Christoph and Webb, Geoffrey I. and Hyndman, Rob J. and Montero-Manso, Pablo\",\n title = \"Monash Time Series Forecasting Archive\",\n booktitle = \"Neural Information Processing Systems Track on Datasets and Benchmarks\",\n year = \"2021\",\n note = \"forthcoming\"\n}\n", "homepage": "https://forecastingdata.org/", "license": "The Creative Commons Attribution 4.0 International License. https://creativecommons.org/licenses/by/4.0/", "features": {"start": {"dtype": "timestamp[s]", "id": null, "_type": "Value"}, "target": {"feature": {"dtype": "float32", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "feat_static_cat": {"feature": {"dtype": "uint64", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "feat_dynamic_real": {"feature": {"feature": {"dtype": "float32", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "length": -1, "id": null, "_type": "Sequence"}, "item_id": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "monash_tsf", "config_name": "hospital", "version": {"version_str": "1.0.0", "description": null, "major": 1, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 217625, "num_examples": 767, "dataset_name": "monash_tsf"}, "test": {"name": "test", "num_bytes": 293558, "num_examples": 767, "dataset_name": "monash_tsf"}, "validation": {"name": "validation", "num_bytes": 255591, "num_examples": 767, "dataset_name": "monash_tsf"}}, "download_checksums": {"https://zenodo.org/record/4656014/files/hospital_dataset.zip": {"num_bytes": 78110, "checksum": "8fd4085efb4517508e30fc69247eee9b74b300765bc6e94482c5d250efba66e5"}}, "download_size": 78110, "post_processing_size": null, "dataset_size": 766774, "size_in_bytes": 844884}, "covid_deaths": {"description": "The first repository containing datasets of related time series for global forecasting.\n", "citation": "@InProceedings{godahewa2021monash,\n author = \"Godahewa, Rakshitha and Bergmeir, Christoph and Webb, Geoffrey I. and Hyndman, Rob J. and Montero-Manso, Pablo\",\n title = \"Monash Time Series Forecasting Archive\",\n booktitle = \"Neural Information Processing Systems Track on Datasets and Benchmarks\",\n year = \"2021\",\n note = \"forthcoming\"\n}\n", "homepage": "https://forecastingdata.org/", "license": "The Creative Commons Attribution 4.0 International License. https://creativecommons.org/licenses/by/4.0/", "features": {"start": {"dtype": "timestamp[s]", "id": null, "_type": "Value"}, "target": {"feature": {"dtype": "float32", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "feat_static_cat": {"feature": {"dtype": "uint64", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "feat_dynamic_real": {"feature": {"feature": {"dtype": "float32", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "length": -1, "id": null, "_type": "Sequence"}, "item_id": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "monash_tsf", "config_name": "covid_deaths", "version": {"version_str": "1.0.0", "description": null, "major": 1, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 176352, "num_examples": 266, "dataset_name": "monash_tsf"}, "test": {"name": "test", "num_bytes": 242187, "num_examples": 266, "dataset_name": "monash_tsf"}, "validation": {"name": "validation", "num_bytes": 209270, "num_examples": 266, "dataset_name": "monash_tsf"}}, "download_checksums": {"https://zenodo.org/record/4656009/files/covid_deaths_dataset.zip": {"num_bytes": 27335, "checksum": "baf11e0982c307f2dae25ff0a6faeb70ac14db8f2dcc71da75ff57563008f1b2"}}, "download_size": 27335, "post_processing_size": null, "dataset_size": 627809, "size_in_bytes": 655144}, "sunspot": {"description": "The first repository containing datasets of related time series for global forecasting.\n", "citation": "@InProceedings{godahewa2021monash,\n author = \"Godahewa, Rakshitha and Bergmeir, Christoph and Webb, Geoffrey I. and Hyndman, Rob J. and Montero-Manso, Pablo\",\n title = \"Monash Time Series Forecasting Archive\",\n booktitle = \"Neural Information Processing Systems Track on Datasets and Benchmarks\",\n year = \"2021\",\n note = \"forthcoming\"\n}\n", "homepage": "https://forecastingdata.org/", "license": "The Creative Commons Attribution 4.0 International License. https://creativecommons.org/licenses/by/4.0/", "features": {"start": {"dtype": "timestamp[s]", "id": null, "_type": "Value"}, "target": {"feature": {"dtype": "float32", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "feat_static_cat": {"feature": {"dtype": "uint64", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "feat_dynamic_real": {"feature": {"feature": {"dtype": "float32", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "length": -1, "id": null, "_type": "Sequence"}, "item_id": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "monash_tsf", "config_name": "sunspot", "version": {"version_str": "1.0.0", "description": null, "major": 1, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 304726, "num_examples": 1, "dataset_name": "monash_tsf"}, "test": {"name": "test", "num_bytes": 304974, "num_examples": 1, "dataset_name": "monash_tsf"}, "validation": {"name": "validation", "num_bytes": 304850, "num_examples": 1, "dataset_name": "monash_tsf"}}, "download_checksums": {"https://zenodo.org/record/4654773/files/sunspot_dataset_with_missing_values.zip": {"num_bytes": 68865, "checksum": "f83d60e56830b30890d0c938dd87e3446658d5fc2b65164b8b3003d8d52ca19e"}}, "download_size": 68865, "post_processing_size": null, "dataset_size": 914550, "size_in_bytes": 983415}, "saugeenday": {"description": "The first repository containing datasets of related time series for global forecasting.\n", "citation": "@InProceedings{godahewa2021monash,\n author = \"Godahewa, Rakshitha and Bergmeir, Christoph and Webb, Geoffrey I. and Hyndman, Rob J. and Montero-Manso, Pablo\",\n title = \"Monash Time Series Forecasting Archive\",\n booktitle = \"Neural Information Processing Systems Track on Datasets and Benchmarks\",\n year = \"2021\",\n note = \"forthcoming\"\n}\n", "homepage": "https://forecastingdata.org/", "license": "The Creative Commons Attribution 4.0 International License. https://creativecommons.org/licenses/by/4.0/", "features": {"start": {"dtype": "timestamp[s]", "id": null, "_type": "Value"}, "target": {"feature": {"dtype": "float32", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "feat_static_cat": {"feature": {"dtype": "uint64", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "feat_dynamic_real": {"feature": {"feature": {"dtype": "float32", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "length": -1, "id": null, "_type": "Sequence"}, "item_id": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "monash_tsf", "config_name": "saugeenday", "version": {"version_str": "1.0.0", "description": null, "major": 1, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 97722, "num_examples": 1, "dataset_name": "monash_tsf"}, "test": {"name": "test", "num_bytes": 97969, "num_examples": 1, "dataset_name": "monash_tsf"}, "validation": {"name": "validation", "num_bytes": 97845, "num_examples": 1, "dataset_name": "monash_tsf"}}, "download_checksums": {"https://zenodo.org/record/4656058/files/saugeenday_dataset.zip": {"num_bytes": 28721, "checksum": "061cb56ee6829d5686e5e08f1a61f4c0af682e32c37824e79926fc11148747df"}}, "download_size": 28721, "post_processing_size": null, "dataset_size": 293536, "size_in_bytes": 322257}, "us_births": {"description": "The first repository containing datasets of related time series for global forecasting.\n", "citation": "@InProceedings{godahewa2021monash,\n author = \"Godahewa, Rakshitha and Bergmeir, Christoph and Webb, Geoffrey I. and Hyndman, Rob J. and Montero-Manso, Pablo\",\n title = \"Monash Time Series Forecasting Archive\",\n booktitle = \"Neural Information Processing Systems Track on Datasets and Benchmarks\",\n year = \"2021\",\n note = \"forthcoming\"\n}\n", "homepage": "https://forecastingdata.org/", "license": "The Creative Commons Attribution 4.0 International License. https://creativecommons.org/licenses/by/4.0/", "features": {"start": {"dtype": "timestamp[s]", "id": null, "_type": "Value"}, "target": {"feature": {"dtype": "float32", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "feat_static_cat": {"feature": {"dtype": "uint64", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "feat_dynamic_real": {"feature": {"feature": {"dtype": "float32", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "length": -1, "id": null, "_type": "Sequence"}, "item_id": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "monash_tsf", "config_name": "us_births", "version": {"version_str": "1.0.0", "description": null, "major": 1, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 29923, "num_examples": 1, "dataset_name": "monash_tsf"}, "test": {"name": "test", "num_bytes": 30171, "num_examples": 1, "dataset_name": "monash_tsf"}, "validation": {"name": "validation", "num_bytes": 30047, "num_examples": 1, "dataset_name": "monash_tsf"}}, "download_checksums": {"https://zenodo.org/record/4656049/files/us_births_dataset.zip": {"num_bytes": 16332, "checksum": "cca9ebce263e6a8cd7c776c002e01ff1f904d8bce5dd2c8103cbed8bc6f8b391"}}, "download_size": 16332, "post_processing_size": null, "dataset_size": 90141, "size_in_bytes": 106473}, "solar_4_seconds": {"description": "The first repository containing datasets of related time series for global forecasting.\n", "citation": "@InProceedings{godahewa2021monash,\n author = \"Godahewa, Rakshitha and Bergmeir, Christoph and Webb, Geoffrey I. and Hyndman, Rob J. and Montero-Manso, Pablo\",\n title = \"Monash Time Series Forecasting Archive\",\n booktitle = \"Neural Information Processing Systems Track on Datasets and Benchmarks\",\n year = \"2021\",\n note = \"forthcoming\"\n}\n", "homepage": "https://forecastingdata.org/", "license": "The Creative Commons Attribution 4.0 International License. https://creativecommons.org/licenses/by/4.0/", "features": {"start": {"dtype": "timestamp[s]", "id": null, "_type": "Value"}, "target": {"feature": {"dtype": "float32", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "feat_static_cat": {"feature": {"dtype": "uint64", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "feat_dynamic_real": {"feature": {"feature": {"dtype": "float32", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "length": -1, "id": null, "_type": "Sequence"}, "item_id": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "monash_tsf", "config_name": "solar_4_seconds", "version": {"version_str": "1.0.0", "description": null, "major": 1, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 30513083, "num_examples": 1, "dataset_name": "monash_tsf"}, "test": {"name": "test", "num_bytes": 30513578, "num_examples": 1, "dataset_name": "monash_tsf"}, "validation": {"name": "validation", "num_bytes": 30513331, "num_examples": 1, "dataset_name": "monash_tsf"}}, "download_checksums": {"https://zenodo.org/record/4656027/files/solar_4_seconds_dataset.zip": {"num_bytes": 794502, "checksum": "9aff580f3cce6b8864572117b2b1cbbf410cc46e9ec40a90201c14bd0296f4fd"}}, "download_size": 794502, "post_processing_size": null, "dataset_size": 91539992, "size_in_bytes": 92334494}, "wind_4_seconds": {"description": "The first repository containing datasets of related time series for global forecasting.\n", "citation": "@InProceedings{godahewa2021monash,\n author = \"Godahewa, Rakshitha and Bergmeir, Christoph and Webb, Geoffrey I. and Hyndman, Rob J. and Montero-Manso, Pablo\",\n title = \"Monash Time Series Forecasting Archive\",\n booktitle = \"Neural Information Processing Systems Track on Datasets and Benchmarks\",\n year = \"2021\",\n note = \"forthcoming\"\n}\n", "homepage": "https://forecastingdata.org/", "license": "The Creative Commons Attribution 4.0 International License. https://creativecommons.org/licenses/by/4.0/", "features": {"start": {"dtype": "timestamp[s]", "id": null, "_type": "Value"}, "target": {"feature": {"dtype": "float32", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "feat_static_cat": {"feature": {"dtype": "uint64", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "feat_dynamic_real": {"feature": {"feature": {"dtype": "float32", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "length": -1, "id": null, "_type": "Sequence"}, "item_id": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "monash_tsf", "config_name": "wind_4_seconds", "version": {"version_str": "1.0.0", "description": null, "major": 1, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 30512774, "num_examples": 1, "dataset_name": "monash_tsf"}, "test": {"name": "test", "num_bytes": 30513269, "num_examples": 1, "dataset_name": "monash_tsf"}, "validation": {"name": "validation", "num_bytes": 30513021, "num_examples": 1, "dataset_name": "monash_tsf"}}, "download_checksums": {"https://zenodo.org/record/4656032/files/wind_4_seconds_dataset.zip": {"num_bytes": 2226184, "checksum": "4214cbf21c1b0f27e4ce06e5ed4fe3b900d49ce81b57c27b1495cf3d81792bca"}}, "download_size": 2226184, "post_processing_size": null, "dataset_size": 91539064, "size_in_bytes": 93765248}, "rideshare": {"description": "The first repository containing datasets of related time series for global forecasting.\n", "citation": "@InProceedings{godahewa2021monash,\n author = \"Godahewa, Rakshitha and Bergmeir, Christoph and Webb, Geoffrey I. and Hyndman, Rob J. and Montero-Manso, Pablo\",\n title = \"Monash Time Series Forecasting Archive\",\n booktitle = \"Neural Information Processing Systems Track on Datasets and Benchmarks\",\n year = \"2021\",\n note = \"forthcoming\"\n}\n", "homepage": "https://forecastingdata.org/", "license": "The Creative Commons Attribution 4.0 International License. https://creativecommons.org/licenses/by/4.0/", "features": {"start": {"dtype": "timestamp[s]", "id": null, "_type": "Value"}, "target": {"feature": {"feature": {"dtype": "float32", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "length": -1, "id": null, "_type": "Sequence"}, "feat_static_cat": {"feature": {"dtype": "uint64", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "feat_dynamic_real": {"feature": {"feature": {"dtype": "float32", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "length": -1, "id": null, "_type": "Sequence"}, "item_id": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "monash_tsf", "config_name": "rideshare", "version": {"version_str": "1.0.0", "description": null, "major": 1, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 4249051, "num_examples": 156, "dataset_name": "monash_tsf"}, "test": {"name": "test", "num_bytes": 5161435, "num_examples": 156, "dataset_name": "monash_tsf"}, "validation": {"name": "validation", "num_bytes": 4705243, "num_examples": 156, "dataset_name": "monash_tsf"}}, "download_checksums": {"https://zenodo.org/record/5122114/files/rideshare_dataset_with_missing_values.zip": {"num_bytes": 1031826, "checksum": "5f83feec4970ff306108f50ac3c13f1adddbf96274226bc0b03655dfdab04567"}}, "download_size": 1031826, "post_processing_size": null, "dataset_size": 14115729, "size_in_bytes": 15147555}, "oikolab_weather": {"description": "The first repository containing datasets of related time series for global forecasting.\n", "citation": "@InProceedings{godahewa2021monash,\n author = \"Godahewa, Rakshitha and Bergmeir, Christoph and Webb, Geoffrey I. and Hyndman, Rob J. and Montero-Manso, Pablo\",\n title = \"Monash Time Series Forecasting Archive\",\n booktitle = \"Neural Information Processing Systems Track on Datasets and Benchmarks\",\n year = \"2021\",\n note = \"forthcoming\"\n}\n", "homepage": "https://forecastingdata.org/", "license": "The Creative Commons Attribution 4.0 International License. https://creativecommons.org/licenses/by/4.0/", "features": {"start": {"dtype": "timestamp[s]", "id": null, "_type": "Value"}, "target": {"feature": {"dtype": "float32", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "feat_static_cat": {"feature": {"dtype": "uint64", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "feat_dynamic_real": {"feature": {"feature": {"dtype": "float32", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "length": -1, "id": null, "_type": "Sequence"}, "item_id": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "monash_tsf", "config_name": "oikolab_weather", "version": {"version_str": "1.0.0", "description": null, "major": 1, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 3299142, "num_examples": 8, "dataset_name": "monash_tsf"}, "test": {"name": "test", "num_bytes": 3302310, "num_examples": 8, "dataset_name": "monash_tsf"}, "validation": {"name": "validation", "num_bytes": 3300726, "num_examples": 8, "dataset_name": "monash_tsf"}}, "download_checksums": {"https://zenodo.org/record/5184708/files/oikolab_weather_dataset.zip": {"num_bytes": 1326101, "checksum": "6f0d2dce3a5aa17c26627ea6107cbf4e5b10050f8a1648d6d03306cd0777f735"}}, "download_size": 1326101, "post_processing_size": null, "dataset_size": 9902178, "size_in_bytes": 11228279}, "temperature_rain": {"description": "The first repository containing datasets of related time series for global forecasting.\n", "citation": "@InProceedings{godahewa2021monash,\n author = \"Godahewa, Rakshitha and Bergmeir, Christoph and Webb, Geoffrey I. and Hyndman, Rob J. and Montero-Manso, Pablo\",\n title = \"Monash Time Series Forecasting Archive\",\n booktitle = \"Neural Information Processing Systems Track on Datasets and Benchmarks\",\n year = \"2021\",\n note = \"forthcoming\"\n}\n", "homepage": "https://forecastingdata.org/", "license": "The Creative Commons Attribution 4.0 International License. https://creativecommons.org/licenses/by/4.0/", "features": {"start": {"dtype": "timestamp[s]", "id": null, "_type": "Value"}, "target": {"feature": {"feature": {"dtype": "float32", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "length": -1, "id": null, "_type": "Sequence"}, "feat_static_cat": {"feature": {"dtype": "uint64", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "feat_dynamic_real": {"feature": {"feature": {"dtype": "float32", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "length": -1, "id": null, "_type": "Sequence"}, "item_id": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "monash_tsf", "config_name": "temperature_rain", "version": {"version_str": "1.0.0", "description": null, "major": 1, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 88121466, "num_examples": 422, "dataset_name": "monash_tsf"}, "test": {"name": "test", "num_bytes": 96059286, "num_examples": 422, "dataset_name": "monash_tsf"}, "validation": {"name": "validation", "num_bytes": 92090376, "num_examples": 422, "dataset_name": "monash_tsf"}}, "download_checksums": {"https://zenodo.org/record/5129073/files/temperature_rain_dataset_with_missing_values.zip": {"num_bytes": 25747139, "checksum": "36a8c8cce2a99a8372ff8f82450b50dcc1b5eff653820c7a828d56730fe325fb"}}, "download_size": 25747139, "post_processing_size": null, "dataset_size": 276271128, "size_in_bytes": 302018267}}
dummy/weather/1.0.0/dummy_data.zip ADDED
@@ -0,0 +1,3 @@
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:992318a4d980d72bc34872ef79fab145f77e675132511745441ff7414af5b89d
3
+ size 41309
monash_tsf.py ADDED
@@ -0,0 +1,566 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Copyright 2020 The HuggingFace Datasets Authors and the current dataset script contributor.
2
+ #
3
+ # Licensed under the Apache License, Version 2.0 (the "License");
4
+ # you may not use this file except in compliance with the License.
5
+ # You may obtain a copy of the License at
6
+ #
7
+ # http://www.apache.org/licenses/LICENSE-2.0
8
+ #
9
+ # Unless required by applicable law or agreed to in writing, software
10
+ # distributed under the License is distributed on an "AS IS" BASIS,
11
+ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
12
+ # See the License for the specific language governing permissions and
13
+ # limitations under the License.
14
+ """Monash Time Series Forecasting Repository Dataset."""
15
+
16
+
17
+ from dataclasses import dataclass
18
+ from datetime import datetime
19
+ from pathlib import Path
20
+ from typing import List, Optional
21
+
22
+ import numpy as np
23
+ from pandas.tseries.frequencies import to_offset
24
+
25
+ import datasets
26
+
27
+ from .utils import convert_tsf_to_dataframe, frequency_converter
28
+
29
+
30
+ # Find for instance the citation on arxiv or on the dataset repo/website
31
+ _CITATION = """\
32
+ @InProceedings{godahewa2021monash,
33
+ author = "Godahewa, Rakshitha and Bergmeir, Christoph and Webb, Geoffrey I. and Hyndman, Rob J. and Montero-Manso, Pablo",
34
+ title = "Monash Time Series Forecasting Archive",
35
+ booktitle = "Neural Information Processing Systems Track on Datasets and Benchmarks",
36
+ year = "2021",
37
+ note = "forthcoming"
38
+ }
39
+ """
40
+
41
+ _DESCRIPTION = """\
42
+ Monash Time Series Forecasting Repository which contains 30+ datasets of related time series for global forecasting research. This repository includes both real-world and competition time series datasets covering varied domains.
43
+ """
44
+
45
+ _HOMEPAGE = "https://forecastingdata.org/"
46
+
47
+ _LICENSE = "The Creative Commons Attribution 4.0 International License. https://creativecommons.org/licenses/by/4.0/"
48
+
49
+ _ROOT_URL = "https://zenodo.org/record"
50
+
51
+
52
+ @dataclass
53
+ class MonashTSFBuilderConfig(datasets.BuilderConfig):
54
+ """MonashTSF builder config with some added meta data."""
55
+
56
+ file_name: Optional[str] = None
57
+ record: Optional[str] = None
58
+ prediction_length: Optional[int] = None
59
+ item_id_column: Optional[str] = None
60
+ data_column: Optional[str] = None
61
+ target_fields: Optional[List[str]] = None
62
+ feat_dynamic_real_fields: Optional[List[str]] = None
63
+ multivariate: bool = False
64
+ rolling_evaluations: int = 1
65
+
66
+
67
+ class MonashTSF(datasets.GeneratorBasedBuilder):
68
+ """Builder of Monash Time Series Forecasting repository of datasets."""
69
+
70
+ VERSION = datasets.Version("1.0.0")
71
+
72
+ BUILDER_CONFIG_CLASS = MonashTSFBuilderConfig
73
+
74
+ BUILDER_CONFIGS = [
75
+ MonashTSFBuilderConfig(
76
+ name="weather",
77
+ version=VERSION,
78
+ description="3010 daily time series representing the variations of four weather variables: rain, mintemp, maxtemp and solar radiation, measured at the weather stations in Australia.",
79
+ record="4654822",
80
+ file_name="weather_dataset.zip",
81
+ data_column="series_type",
82
+ ),
83
+ MonashTSFBuilderConfig(
84
+ name="tourism_yearly",
85
+ version=VERSION,
86
+ description="This dataset contains 518 yearly time series used in the Kaggle Tourism forecasting competition.",
87
+ record="4656103",
88
+ file_name="tourism_yearly_dataset.zip",
89
+ ),
90
+ MonashTSFBuilderConfig(
91
+ name="tourism_quarterly",
92
+ version=VERSION,
93
+ description="This dataset contains 427 quarterly time series used in the Kaggle Tourism forecasting competition.",
94
+ record="4656093",
95
+ file_name="tourism_quarterly_dataset.zip",
96
+ ),
97
+ MonashTSFBuilderConfig(
98
+ name="tourism_monthly",
99
+ version=VERSION,
100
+ description="This dataset contains 366 monthly time series used in the Kaggle Tourism forecasting competition.",
101
+ record="4656096",
102
+ file_name="tourism_monthly_dataset.zip",
103
+ ),
104
+ MonashTSFBuilderConfig(
105
+ name="cif_2016",
106
+ version=VERSION,
107
+ description="72 monthly time series originated from the banking domain used in the CIF 2016 forecasting competition.",
108
+ record="4656042",
109
+ file_name="cif_2016_dataset.zip",
110
+ ),
111
+ MonashTSFBuilderConfig(
112
+ name="london_smart_meters",
113
+ version=VERSION,
114
+ description="5560 half hourly time series that represent the energy consumption readings of London households in kilowatt hour (kWh) from November 2011 to February 2014.",
115
+ record="4656072",
116
+ file_name="london_smart_meters_dataset_with_missing_values.zip",
117
+ ),
118
+ MonashTSFBuilderConfig(
119
+ name="australian_electricity_demand",
120
+ version=VERSION,
121
+ description="5 time series representing the half hourly electricity demand of 5 states in Australia: Victoria, New South Wales, Queensland, Tasmania and South Australia.",
122
+ record="4659727",
123
+ file_name="australian_electricity_demand_dataset.zip",
124
+ ),
125
+ MonashTSFBuilderConfig(
126
+ name="wind_farms_minutely",
127
+ version=VERSION,
128
+ description="Minutely time series representing the wind power production of 339 wind farms in Australia.",
129
+ record="4654909",
130
+ file_name="wind_farms_minutely_dataset_with_missing_values.zip",
131
+ ),
132
+ MonashTSFBuilderConfig(
133
+ name="bitcoin",
134
+ version=VERSION,
135
+ description="18 daily time series including hash rate, block size, mining difficulty etc. as well as public opinion in the form of tweets and google searches mentioning the keyword bitcoin as potential influencer of the bitcoin price.",
136
+ record="5121965",
137
+ file_name="bitcoin_dataset_with_missing_values.zip",
138
+ ),
139
+ MonashTSFBuilderConfig(
140
+ name="pedestrian_counts",
141
+ version=VERSION,
142
+ description="Hourly pedestrian counts captured from 66 sensors in Melbourne city starting from May 2009.",
143
+ record="4656626",
144
+ file_name="pedestrian_counts_dataset.zip",
145
+ ),
146
+ MonashTSFBuilderConfig(
147
+ name="vehicle_trips",
148
+ version=VERSION,
149
+ description="329 daily time series representing the number of trips and vehicles belonging to a set of for-hire vehicle (FHV) companies.",
150
+ record="5122535",
151
+ file_name="vehicle_trips_dataset_with_missing_values.zip",
152
+ ),
153
+ MonashTSFBuilderConfig(
154
+ name="kdd_cup_2018",
155
+ version=VERSION,
156
+ description="Hourly time series representing the air quality levels in 59 stations in 2 cities from 01/01/2017 to 31/03/2018.",
157
+ record="4656719",
158
+ file_name="kdd_cup_2018_dataset_with_missing_values.zip",
159
+ ),
160
+ MonashTSFBuilderConfig(
161
+ name="nn5_daily",
162
+ version=VERSION,
163
+ description="111 time series to predicting the daily cash withdrawals from ATMs in UK.",
164
+ record="4656110",
165
+ file_name="nn5_daily_dataset_with_missing_values.zip",
166
+ ),
167
+ MonashTSFBuilderConfig(
168
+ name="nn5_weekly",
169
+ version=VERSION,
170
+ description="111 time series to predicting the weekly cash withdrawals from ATMs in UK.",
171
+ record="4656125",
172
+ file_name="nn5_weekly_dataset.zip",
173
+ ),
174
+ MonashTSFBuilderConfig(
175
+ name="kaggle_web_traffic",
176
+ version=VERSION,
177
+ description="145063 daily time series representing the number of hits or web traffic for a set of Wikipedia pages from 2015-07-01 to 2017-09-10.",
178
+ record="4656080",
179
+ file_name="kaggle_web_traffic_dataset_with_missing_values.zip",
180
+ ),
181
+ MonashTSFBuilderConfig(
182
+ name="kaggle_web_traffic_weekly",
183
+ version=VERSION,
184
+ description="145063 daily time series representing the number of hits or web traffic for a set of Wikipedia pages from 2015-07-01 to 2017-09-10.",
185
+ record="4656664",
186
+ file_name="kaggle_web_traffic_weekly_dataset.zip",
187
+ ),
188
+ MonashTSFBuilderConfig(
189
+ name="solar_10_minutes",
190
+ version=VERSION,
191
+ description="137 time series representing the solar power production recorded per every 10 minutes in Alabama state in 2006.",
192
+ record="4656144",
193
+ file_name="solar_10_minutes_dataset.zip",
194
+ ),
195
+ MonashTSFBuilderConfig(
196
+ name="solar_weekly",
197
+ version=VERSION,
198
+ description="137 time series representing the weekly solar power production in Alabama state in 2006.",
199
+ record="4656151",
200
+ file_name="solar_weekly_dataset.zip",
201
+ ),
202
+ MonashTSFBuilderConfig(
203
+ name="car_parts",
204
+ version=VERSION,
205
+ description="2674 intermittent monthly time series that represent car parts sales from January 1998 to March 2002.",
206
+ record="4656022",
207
+ file_name="car_parts_dataset_with_missing_values.zip",
208
+ ),
209
+ MonashTSFBuilderConfig(
210
+ name="fred_md",
211
+ version=VERSION,
212
+ description="107 monthly time series showing a set of macro-economic indicators from the Federal Reserve Bank.",
213
+ record="4654833",
214
+ file_name="fred_md_dataset.zip",
215
+ ),
216
+ MonashTSFBuilderConfig(
217
+ name="traffic_hourly",
218
+ version=VERSION,
219
+ description="862 hourly time series showing the road occupancy rates on the San Francisco Bay area freeways from 2015 to 2016.",
220
+ record="4656132",
221
+ file_name="traffic_hourly_dataset.zip",
222
+ ),
223
+ MonashTSFBuilderConfig(
224
+ name="traffic_weekly",
225
+ version=VERSION,
226
+ description="862 weekly time series showing the road occupancy rates on the San Francisco Bay area freeways from 2015 to 2016.",
227
+ record="4656135",
228
+ file_name="traffic_weekly_dataset.zip",
229
+ ),
230
+ MonashTSFBuilderConfig(
231
+ name="hospital",
232
+ version=VERSION,
233
+ description="767 monthly time series that represent the patient counts related to medical products from January 2000 to December 2006.",
234
+ record="4656014",
235
+ file_name="hospital_dataset.zip",
236
+ ),
237
+ MonashTSFBuilderConfig(
238
+ name="covid_deaths",
239
+ version=VERSION,
240
+ description="266 daily time series that represent the COVID-19 deaths in a set of countries and states from 22/01/2020 to 20/08/2020.",
241
+ record="4656009",
242
+ file_name="covid_deaths_dataset.zip",
243
+ ),
244
+ MonashTSFBuilderConfig(
245
+ name="sunspot",
246
+ version=VERSION,
247
+ description="A single very long daily time series of sunspot numbers from 1818-01-08 to 2020-05-31.",
248
+ record="4654773",
249
+ file_name="sunspot_dataset_with_missing_values.zip",
250
+ ),
251
+ MonashTSFBuilderConfig(
252
+ name="saugeenday",
253
+ version=VERSION,
254
+ description="A single very long time series representing the daily mean flow of the Saugeen River at Walkerton in cubic meters per second from 01/01/1915 to 31/12/1979.",
255
+ record="4656058",
256
+ file_name="saugeenday_dataset.zip",
257
+ ),
258
+ MonashTSFBuilderConfig(
259
+ name="us_births",
260
+ version=VERSION,
261
+ description="A single very long daily time series representing the number of births in US from 01/01/1969 to 31/12/1988.",
262
+ record="4656049",
263
+ file_name="us_births_dataset.zip",
264
+ ),
265
+ MonashTSFBuilderConfig(
266
+ name="solar_4_seconds",
267
+ version=VERSION,
268
+ description="A single very long daily time series representing the solar power production in MW recorded per every 4 seconds starting from 01/08/2019.",
269
+ record="4656027",
270
+ file_name="solar_4_seconds_dataset.zip",
271
+ ),
272
+ MonashTSFBuilderConfig(
273
+ name="wind_4_seconds",
274
+ version=VERSION,
275
+ description="A single very long daily time series representing the wind power production in MW recorded per every 4 seconds starting from 01/08/2019.",
276
+ record="4656032",
277
+ file_name="wind_4_seconds_dataset.zip",
278
+ ),
279
+ MonashTSFBuilderConfig(
280
+ name="rideshare",
281
+ version=VERSION,
282
+ description="156 hourly time series representations of attributes related to Uber and Lyft rideshare services for various locations in New York between 26/11/2018 and 18/12/2018.",
283
+ record="5122114",
284
+ file_name="rideshare_dataset_with_missing_values.zip",
285
+ item_id_column=["source_location", "provider_name", "provider_service"],
286
+ data_column="type",
287
+ target_fields=[
288
+ "price_min",
289
+ "price_mean",
290
+ "price_max",
291
+ "distance_min",
292
+ "distance_mean",
293
+ "distance_max",
294
+ "surge_min",
295
+ "surge_mean",
296
+ "surge_max",
297
+ "api_calls",
298
+ ],
299
+ feat_dynamic_real_fields=["temp", "rain", "humidity", "clouds", "wind"],
300
+ multivariate=True,
301
+ ),
302
+ MonashTSFBuilderConfig(
303
+ name="oikolab_weather",
304
+ version=VERSION,
305
+ description="Eight time series representing the hourly climate data nearby Monash University, Clayton, Victoria, Australia from 2010-01-01 to 2021-05-31",
306
+ record="5184708",
307
+ file_name="oikolab_weather_dataset.zip",
308
+ data_column="type",
309
+ ),
310
+ MonashTSFBuilderConfig(
311
+ name="temperature_rain",
312
+ version=VERSION,
313
+ description="32072 daily time series showing the temperature observations and rain forecasts, gathered by the Australian Bureau of Meteorology for 422 weather stations across Australia, between 02/05/2015 and 26/04/2017",
314
+ record="5129073",
315
+ file_name="temperature_rain_dataset_with_missing_values.zip",
316
+ item_id_column="station_id",
317
+ data_column="obs_or_fcst",
318
+ target_fields=[
319
+ "fcst_0_DailyPoP",
320
+ "fcst_0_DailyPoP1",
321
+ "fcst_0_DailyPoP10",
322
+ "fcst_0_DailyPoP15",
323
+ "fcst_0_DailyPoP25",
324
+ "fcst_0_DailyPoP5",
325
+ "fcst_0_DailyPoP50",
326
+ "fcst_0_DailyPrecip",
327
+ "fcst_0_DailyPrecip10Pct",
328
+ "fcst_0_DailyPrecip25Pct",
329
+ "fcst_0_DailyPrecip50Pct",
330
+ "fcst_0_DailyPrecip75Pct",
331
+ "fcst_1_DailyPoP",
332
+ "fcst_1_DailyPoP1",
333
+ "fcst_1_DailyPoP10",
334
+ "fcst_1_DailyPoP15",
335
+ "fcst_1_DailyPoP25",
336
+ "fcst_1_DailyPoP5",
337
+ "fcst_1_DailyPoP50",
338
+ "fcst_1_DailyPrecip",
339
+ "fcst_1_DailyPrecip10Pct",
340
+ "fcst_1_DailyPrecip25Pct",
341
+ "fcst_1_DailyPrecip50Pct",
342
+ "fcst_1_DailyPrecip75Pct",
343
+ "fcst_2_DailyPoP",
344
+ "fcst_2_DailyPoP1",
345
+ "fcst_2_DailyPoP10",
346
+ "fcst_2_DailyPoP15",
347
+ "fcst_2_DailyPoP25",
348
+ "fcst_2_DailyPoP5",
349
+ "fcst_2_DailyPoP50",
350
+ "fcst_2_DailyPrecip",
351
+ "fcst_2_DailyPrecip10Pct",
352
+ "fcst_2_DailyPrecip25Pct",
353
+ "fcst_2_DailyPrecip50Pct",
354
+ "fcst_2_DailyPrecip75Pct",
355
+ "fcst_3_DailyPoP",
356
+ "fcst_3_DailyPoP1",
357
+ "fcst_3_DailyPoP10",
358
+ "fcst_3_DailyPoP15",
359
+ "fcst_3_DailyPoP25",
360
+ "fcst_3_DailyPoP5",
361
+ "fcst_3_DailyPoP50",
362
+ "fcst_3_DailyPrecip",
363
+ "fcst_3_DailyPrecip10Pct",
364
+ "fcst_3_DailyPrecip25Pct",
365
+ "fcst_3_DailyPrecip50Pct",
366
+ "fcst_3_DailyPrecip75Pct",
367
+ "fcst_4_DailyPoP",
368
+ "fcst_4_DailyPoP1",
369
+ "fcst_4_DailyPoP10",
370
+ "fcst_4_DailyPoP15",
371
+ "fcst_4_DailyPoP25",
372
+ "fcst_4_DailyPoP5",
373
+ "fcst_4_DailyPoP50",
374
+ "fcst_4_DailyPrecip",
375
+ "fcst_4_DailyPrecip10Pct",
376
+ "fcst_4_DailyPrecip25Pct",
377
+ "fcst_4_DailyPrecip50Pct",
378
+ "fcst_4_DailyPrecip75Pct",
379
+ "fcst_5_DailyPoP",
380
+ "fcst_5_DailyPoP1",
381
+ "fcst_5_DailyPoP10",
382
+ "fcst_5_DailyPoP15",
383
+ "fcst_5_DailyPoP25",
384
+ "fcst_5_DailyPoP5",
385
+ "fcst_5_DailyPoP50",
386
+ "fcst_5_DailyPrecip",
387
+ "fcst_5_DailyPrecip10Pct",
388
+ "fcst_5_DailyPrecip25Pct",
389
+ "fcst_5_DailyPrecip50Pct",
390
+ "fcst_5_DailyPrecip75Pct",
391
+ ],
392
+ feat_dynamic_real_fields=["T_MEAN", "PRCP_SUM", "T_MAX", "T_MIN"],
393
+ multivariate=True,
394
+ ),
395
+ ]
396
+
397
+ def _info(self):
398
+ if self.config.multivariate:
399
+ features = datasets.Features(
400
+ {
401
+ "start": datasets.Value("timestamp[s]"),
402
+ "target": datasets.Sequence(datasets.Sequence(datasets.Value("float32"))),
403
+ "feat_static_cat": datasets.Sequence(datasets.Value("uint64")),
404
+ # "feat_static_real": datasets.Sequence(datasets.Value("float32")),
405
+ "feat_dynamic_real": datasets.Sequence(datasets.Sequence(datasets.Value("float32"))),
406
+ # "feat_dynamic_cat": datasets.Sequence(datasets.Sequence(datasets.Value("uint64"))),
407
+ "item_id": datasets.Value("string"),
408
+ }
409
+ )
410
+ else:
411
+ features = datasets.Features(
412
+ {
413
+ "start": datasets.Value("timestamp[s]"),
414
+ "target": datasets.Sequence(datasets.Value("float32")),
415
+ "feat_static_cat": datasets.Sequence(datasets.Value("uint64")),
416
+ # "feat_static_real": datasets.Sequence(datasets.Value("float32")),
417
+ "feat_dynamic_real": datasets.Sequence(datasets.Sequence(datasets.Value("float32"))),
418
+ # "feat_dynamic_cat": datasets.Sequence(datasets.Sequence(datasets.Value("uint64"))),
419
+ "item_id": datasets.Value("string"),
420
+ }
421
+ )
422
+
423
+ return datasets.DatasetInfo(
424
+ description=_DESCRIPTION,
425
+ features=features,
426
+ homepage=_HOMEPAGE,
427
+ license=_LICENSE,
428
+ citation=_CITATION,
429
+ )
430
+
431
+ def _split_generators(self, dl_manager):
432
+ # dl_manager is a datasets.download.DownloadManager that can be used to download and extract URLS
433
+ # It can accept any type or nested list/dict and will give back the same structure with the url replaced with path to local files.
434
+ # By default the archives will be extracted and a path to a cached folder where they are extracted is returned instead of the archive
435
+ urls = f"{_ROOT_URL}/{self.config.record}/files/{self.config.file_name}"
436
+ data_dir = dl_manager.download_and_extract(urls)
437
+ file_path = Path(data_dir) / (self.config.file_name.split(".")[0] + ".tsf")
438
+
439
+ return [
440
+ datasets.SplitGenerator(
441
+ name=datasets.Split.TRAIN,
442
+ # These kwargs will be passed to _generate_examples
443
+ gen_kwargs={
444
+ "filepath": file_path,
445
+ "split": "train",
446
+ },
447
+ ),
448
+ datasets.SplitGenerator(
449
+ name=datasets.Split.TEST,
450
+ # These kwargs will be passed to _generate_examples
451
+ gen_kwargs={"filepath": file_path, "split": "test"},
452
+ ),
453
+ datasets.SplitGenerator(
454
+ name=datasets.Split.VALIDATION,
455
+ # These kwargs will be passed to _generate_examples
456
+ gen_kwargs={
457
+ "filepath": file_path,
458
+ "split": "val",
459
+ },
460
+ ),
461
+ ]
462
+
463
+ # method parameters are unpacked from `gen_kwargs` as given in `_split_generators`
464
+ def _generate_examples(self, filepath, split):
465
+ (
466
+ loaded_data,
467
+ frequency,
468
+ forecast_horizon,
469
+ _,
470
+ _,
471
+ ) = convert_tsf_to_dataframe(filepath, value_column_name="target")
472
+
473
+ if forecast_horizon is None:
474
+ prediction_length_map = {
475
+ "S": 60,
476
+ "T": 60,
477
+ "H": 48,
478
+ "D": 30,
479
+ "W": 8,
480
+ "M": 12,
481
+ "Y": 4,
482
+ }
483
+ freq = frequency_converter(frequency)
484
+ freq = to_offset(freq).name
485
+ forecast_horizon = prediction_length_map[freq]
486
+
487
+ if self.config.prediction_length is not None:
488
+ forecast_horizon = self.config.prediction_length
489
+
490
+ if self.config.item_id_column is not None:
491
+ loaded_data.set_index(self.config.item_id_column, inplace=True)
492
+ loaded_data.sort_index(inplace=True)
493
+
494
+ for cat, item_id in enumerate(loaded_data.index.unique()):
495
+ ts = loaded_data.loc[item_id]
496
+ start = ts.start_timestamp[0]
497
+
498
+ if self.config.target_fields is not None:
499
+ target_fields = ts[ts[self.config.data_column].isin(self.config.target_fields)]
500
+ else:
501
+ target_fields = self.config.data_column.unique()
502
+
503
+ if self.config.feat_dynamic_real_fields is not None:
504
+ feat_dynamic_real_fields = ts[
505
+ ts[self.config.data_column].isin(self.config.feat_dynamic_real_fields)
506
+ ]
507
+ feat_dynamic_real = np.vstack(feat_dynamic_real_fields.target)
508
+ else:
509
+ feat_dynamic_real = None
510
+
511
+ target = np.vstack(target_fields.target)
512
+
513
+ feat_static_cat = [cat]
514
+
515
+ if split in ["train", "val"]:
516
+ offset = forecast_horizon * self.config.rolling_evaluations + forecast_horizon * (split == "train")
517
+ target = target[..., :-offset]
518
+ if self.config.feat_dynamic_real_fields is not None:
519
+ feat_dynamic_real = feat_dynamic_real[..., :-offset]
520
+
521
+ yield cat, {
522
+ "start": start,
523
+ "target": target,
524
+ "feat_dynamic_real": feat_dynamic_real,
525
+ "feat_static_cat": feat_static_cat,
526
+ "item_id": item_id,
527
+ }
528
+ else:
529
+ if self.config.target_fields is not None:
530
+ target_fields = loaded_data[loaded_data[self.config.data_column].isin(self.config.target_fields)]
531
+ else:
532
+ target_fields = loaded_data
533
+ if self.config.feat_dynamic_real_fields is not None:
534
+ feat_dynamic_real_fields = loaded_data[
535
+ loaded_data[self.config.data_column].isin(self.config.feat_dynamic_real_fields)
536
+ ]
537
+ else:
538
+ feat_dynamic_real_fields = None
539
+
540
+ for cat, ts in target_fields.iterrows():
541
+ start = ts.get("start_timestamp", datetime.strptime("1900-01-01 00-00-00", "%Y-%m-%d %H-%M-%S"))
542
+ target = ts.target
543
+ if feat_dynamic_real_fields is not None:
544
+ feat_dynamic_real = np.vstack(feat_dynamic_real_fields.target)
545
+ else:
546
+ feat_dynamic_real = None
547
+
548
+ feat_static_cat = [cat]
549
+ if self.config.data_column is not None:
550
+ item_id = f"{ts.series_name}-{ts[self.config.data_column]}"
551
+ else:
552
+ item_id = ts.series_name
553
+
554
+ if split in ["train", "val"]:
555
+ offset = forecast_horizon * self.config.rolling_evaluations + forecast_horizon * (split == "train")
556
+ target = target[..., :-offset]
557
+ if feat_dynamic_real is not None:
558
+ feat_dynamic_real = feat_dynamic_real[..., :-offset]
559
+
560
+ yield cat, {
561
+ "start": start,
562
+ "target": target,
563
+ "feat_dynamic_real": feat_dynamic_real,
564
+ "feat_static_cat": feat_static_cat,
565
+ "item_id": item_id,
566
+ }
utils.py ADDED
@@ -0,0 +1,188 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ from datetime import datetime
2
+ from distutils.util import strtobool
3
+
4
+ import numpy as np
5
+ import pandas as pd
6
+
7
+
8
+ # Converts the contents in a .tsf file into a dataframe and returns
9
+ # it along with other meta-data of the dataset:
10
+ # frequency, horizon, whether the dataset contains missing values and whether the series have equal lengths
11
+ #
12
+ # Parameters
13
+ # full_file_path_and_name - complete .tsf file path
14
+ # replace_missing_vals_with - a term to indicate the missing values in series in the returning dataframe
15
+ # value_column_name - Any name that is preferred to have as the name of the column containing series values in the returning dataframe
16
+ def convert_tsf_to_dataframe(
17
+ full_file_path_and_name,
18
+ replace_missing_vals_with="NaN",
19
+ value_column_name="series_value",
20
+ ):
21
+ col_names = []
22
+ col_types = []
23
+ all_data = {}
24
+ line_count = 0
25
+ frequency = None
26
+ forecast_horizon = None
27
+ contain_missing_values = None
28
+ contain_equal_length = None
29
+ found_data_tag = False
30
+ found_data_section = False
31
+ started_reading_data_section = False
32
+
33
+ with open(full_file_path_and_name, "r", encoding="cp1252") as file:
34
+ for line in file:
35
+ # Strip white space from start/end of line
36
+ line = line.strip()
37
+
38
+ if line:
39
+ if line.startswith("@"): # Read meta-data
40
+ if not line.startswith("@data"):
41
+ line_content = line.split(" ")
42
+ if line.startswith("@attribute"):
43
+ if len(line_content) != 3: # Attributes have both name and type
44
+ raise ValueError("Invalid meta-data specification.")
45
+
46
+ col_names.append(line_content[1])
47
+ col_types.append(line_content[2])
48
+ else:
49
+ if len(line_content) != 2: # Other meta-data have only values
50
+ raise ValueError("Invalid meta-data specification.")
51
+
52
+ if line.startswith("@frequency"):
53
+ frequency = line_content[1]
54
+ elif line.startswith("@horizon"):
55
+ forecast_horizon = int(line_content[1])
56
+ elif line.startswith("@missing"):
57
+ contain_missing_values = bool(strtobool(line_content[1]))
58
+ elif line.startswith("@equallength"):
59
+ contain_equal_length = bool(strtobool(line_content[1]))
60
+
61
+ else:
62
+ if len(col_names) == 0:
63
+ raise ValueError("Missing attribute section. Attribute section must come before data.")
64
+
65
+ found_data_tag = True
66
+ elif not line.startswith("#"):
67
+ if len(col_names) == 0:
68
+ raise ValueError("Missing attribute section. Attribute section must come before data.")
69
+ elif not found_data_tag:
70
+ raise ValueError("Missing @data tag.")
71
+ else:
72
+ if not started_reading_data_section:
73
+ started_reading_data_section = True
74
+ found_data_section = True
75
+ all_series = []
76
+
77
+ for col in col_names:
78
+ all_data[col] = []
79
+
80
+ full_info = line.split(":")
81
+
82
+ if len(full_info) != (len(col_names) + 1):
83
+ raise ValueError("Missing attributes/values in series.")
84
+
85
+ series = full_info[len(full_info) - 1]
86
+ series = series.split(",")
87
+
88
+ if len(series) == 0:
89
+ raise ValueError(
90
+ "A given series should contains a set of comma separated numeric values. At least one numeric value should be there in a series. Missing values should be indicated with ? symbol"
91
+ )
92
+
93
+ numeric_series = []
94
+
95
+ for val in series:
96
+ if val == "?":
97
+ numeric_series.append(replace_missing_vals_with)
98
+ else:
99
+ numeric_series.append(float(val))
100
+
101
+ if numeric_series.count(replace_missing_vals_with) == len(numeric_series):
102
+ raise ValueError(
103
+ "All series values are missing. A given series should contains a set of comma separated numeric values. At least one numeric value should be there in a series."
104
+ )
105
+
106
+ all_series.append(np.array(numeric_series, dtype=np.float32))
107
+
108
+ for i in range(len(col_names)):
109
+ att_val = None
110
+ if col_types[i] == "numeric":
111
+ att_val = int(full_info[i])
112
+ elif col_types[i] == "string":
113
+ att_val = str(full_info[i])
114
+ elif col_types[i] == "date":
115
+ att_val = datetime.strptime(full_info[i], "%Y-%m-%d %H-%M-%S")
116
+ else:
117
+ raise ValueError(
118
+ "Invalid attribute type."
119
+ ) # Currently, the code supports only numeric, string and date types. Extend this as required.
120
+
121
+ if att_val is None:
122
+ raise ValueError("Invalid attribute value.")
123
+ else:
124
+ all_data[col_names[i]].append(att_val)
125
+
126
+ line_count = line_count + 1
127
+
128
+ if line_count == 0:
129
+ raise ValueError("Empty file.")
130
+ if len(col_names) == 0:
131
+ raise ValueError("Missing attribute section.")
132
+ if not found_data_section:
133
+ raise ValueError("Missing series information under data section.")
134
+
135
+ all_data[value_column_name] = all_series
136
+ loaded_data = pd.DataFrame(all_data)
137
+
138
+ return (
139
+ loaded_data,
140
+ frequency,
141
+ forecast_horizon,
142
+ contain_missing_values,
143
+ contain_equal_length,
144
+ )
145
+
146
+
147
+ def convert_multiple(text: str) -> str:
148
+ if text.isnumeric():
149
+ return text
150
+ if text == "half":
151
+ return "0.5"
152
+
153
+
154
+ def frequency_converter(freq: str):
155
+ parts = freq.split("_")
156
+ if len(parts) == 1:
157
+ return BASE_FREQ_TO_PANDAS_OFFSET[parts[0]]
158
+ if len(parts) == 2:
159
+ return convert_multiple(parts[0]) + BASE_FREQ_TO_PANDAS_OFFSET[parts[1]]
160
+ raise ValueError(f"Invalid frequency string {freq}.")
161
+
162
+
163
+ BASE_FREQ_TO_PANDAS_OFFSET = {
164
+ "seconds": "S",
165
+ "minutely": "T",
166
+ "minutes": "T",
167
+ "hourly": "H",
168
+ "hours": "H",
169
+ "daily": "D",
170
+ "days": "D",
171
+ "weekly": "W",
172
+ "weeks": "W",
173
+ "monthly": "M",
174
+ "months": "M",
175
+ "quarterly": "Q",
176
+ "quarters": "Q",
177
+ "yearly": "Y",
178
+ "years": "Y",
179
+ }
180
+
181
+ # Example of usage
182
+ # loaded_data, frequency, forecast_horizon, contain_missing_values, contain_equal_length = convert_tsf_to_dataframe("TSForecasting/tsf_data/sample.tsf")
183
+
184
+ # print(loaded_data)
185
+ # print(frequency)
186
+ # print(forecast_horizon)
187
+ # print(contain_missing_values)
188
+ # print(contain_equal_length)