jorses
commited on
Commit
·
b94798c
1
Parent(s):
dedccd2
Update README.md
Browse files
README.md
CHANGED
|
@@ -91,7 +91,7 @@ By clicking on each name in the table below, you will be able to explore each da
|
|
| 91 |
| 64 | [Clustering Zoo Animals](https://public.graphext.com/d1b66902e46a712a/index.html) | 101 | 18 | Health | [Kaggle](https://www.kaggle.com/datasets/jirkadaberger/zoo-animals) |
|
| 92 |
| 65 | [RFM Analysis](https://public.graphext.com/4db2e54e29006a21/index.html) | 541909 | 8 | Business | [UCI ML](https://www.kaggle.com/datasets/carrie1/ecommerce-data) |
|
| 93 |
|
| 94 |
-
## Folder structure
|
| 95 |
Each folder represents one dataset. You will find the following files within:
|
| 96 |
|
| 97 |
* all.parquet: the processed data, with each column tagged with our typing system, in [parquet](https://arrow.apache.org/docs/python/parquet.html).
|
|
@@ -99,7 +99,7 @@ Each folder represents one dataset. You will find the following files within:
|
|
| 99 |
* sample.csv: sample containing 20 rows of the original dataset
|
| 100 |
* info.yml: additional information about the dataset
|
| 101 |
|
| 102 |
-
## Column typing system
|
| 103 |
In an effort to map the stage for later analysis, we have categorized the columns by type. This information allows us to segment different kinds of data so that we can subsequently analyze the model's behavior on each column type separately. All parquet files have been casted to their smallest viable data type using the open source [Lector](https://github.com/graphext/lector) reader.
|
| 104 |
|
| 105 |
What this means is that in the data types we have more granular information that allows us to know if the column contains NaNs or not (following panda’s convention of Int vs int), as well as whether small numerical values contain negatives (Uint vs int) and their range. We also have dates with potential timezone information (although for now they’re all UTC), as well as information about categories’ cardinality coming from the arrow types.
|
|
|
|
| 91 |
| 64 | [Clustering Zoo Animals](https://public.graphext.com/d1b66902e46a712a/index.html) | 101 | 18 | Health | [Kaggle](https://www.kaggle.com/datasets/jirkadaberger/zoo-animals) |
|
| 92 |
| 65 | [RFM Analysis](https://public.graphext.com/4db2e54e29006a21/index.html) | 541909 | 8 | Business | [UCI ML](https://www.kaggle.com/datasets/carrie1/ecommerce-data) |
|
| 93 |
|
| 94 |
+
## 🏗️ Folder structure
|
| 95 |
Each folder represents one dataset. You will find the following files within:
|
| 96 |
|
| 97 |
* all.parquet: the processed data, with each column tagged with our typing system, in [parquet](https://arrow.apache.org/docs/python/parquet.html).
|
|
|
|
| 99 |
* sample.csv: sample containing 20 rows of the original dataset
|
| 100 |
* info.yml: additional information about the dataset
|
| 101 |
|
| 102 |
+
## 🗂️ Column typing system
|
| 103 |
In an effort to map the stage for later analysis, we have categorized the columns by type. This information allows us to segment different kinds of data so that we can subsequently analyze the model's behavior on each column type separately. All parquet files have been casted to their smallest viable data type using the open source [Lector](https://github.com/graphext/lector) reader.
|
| 104 |
|
| 105 |
What this means is that in the data types we have more granular information that allows us to know if the column contains NaNs or not (following panda’s convention of Int vs int), as well as whether small numerical values contain negatives (Uint vs int) and their range. We also have dates with potential timezone information (although for now they’re all UTC), as well as information about categories’ cardinality coming from the arrow types.
|