Datasets:
The dataset viewer is not available for this split.
Error code: FeaturesError
Exception: ArrowInvalid
Message: JSON parse error: Invalid value. in row 0
Traceback: Traceback (most recent call last):
File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/json/json.py", line 324, in _generate_tables
df = pandas_read_json(f)
File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/json/json.py", line 38, in pandas_read_json
return pd.read_json(path_or_buf, **kwargs)
~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^
File "/usr/local/lib/python3.14/site-packages/pandas/io/json/_json.py", line 791, in read_json
json_reader = JsonReader(
path_or_buf,
...<16 lines>...
engine=engine,
)
File "/usr/local/lib/python3.14/site-packages/pandas/io/json/_json.py", line 905, in __init__
self.data = self._preprocess_data(data)
~~~~~~~~~~~~~~~~~~~~~^^^^^^
File "/usr/local/lib/python3.14/site-packages/pandas/io/json/_json.py", line 917, in _preprocess_data
data = data.read()
File "/usr/local/lib/python3.14/site-packages/datasets/utils/file_utils.py", line 839, in read_with_retries
out = read(*args, **kwargs)
File "<frozen codecs>", line 325, in decode
UnicodeDecodeError: 'utf-8' codec can't decode byte 0xac in position 10: invalid start byte
During handling of the above exception, another exception occurred:
Traceback (most recent call last):
File "/src/services/worker/src/worker/job_runners/split/first_rows.py", line 244, in compute_first_rows_from_streaming_response
iterable_dataset = iterable_dataset._resolve_features()
File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 4523, in _resolve_features
features = _infer_features_from_batch(self.with_format(None)._head())
~~~~~~~~~~~~~~~~~~~~~~~~~~~~^^
File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2768, in _head
return next(iter(self.iter(batch_size=n)))
File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2972, in iter
for key, pa_table in ex_iterable.iter_arrow():
~~~~~~~~~~~~~~~~~~~~~~^^
File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2483, in _iter_arrow
yield from self.ex_iterable._iter_arrow()
File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 547, in _iter_arrow
for key, pa_table in iterator:
^^^^^^^^
File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 430, in _iter_arrow
for key, pa_table in self.generate_tables_fn(**gen_kwags):
~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^
File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/json/json.py", line 327, in _generate_tables
raise e
File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/json/json.py", line 290, in _generate_tables
pa_table = paj.read_json(
io.BytesIO(batch), read_options=paj.ReadOptions(block_size=block_size)
)
File "pyarrow/_json.pyx", line 364, in pyarrow._json.read_json
File "pyarrow/error.pxi", line 155, in pyarrow.lib.pyarrow_internal_check_status
File "pyarrow/error.pxi", line 92, in pyarrow.lib.check_status
raise convert_status(status)
pyarrow.lib.ArrowInvalid: JSON parse error: Invalid value. in row 0Need help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.
LEG-SEIR dengue data and saved experiment
This dataset supports the LEG-SEIR code and paper. LEG-SEIR uses three recent reports to predict dengue counts in 25 Sri Lankan districts for the next three reports.
The release contains the checked input tables and the saved outputs from 265 neural training runs. Hugging Face hosts our copy. The original case reports come from Sri Lanka's Epidemiology Unit, and population values come from the Department of Census and Statistics.
Start here
- Read how the model works.
- Use the glossary for unfamiliar terms.
- Follow the run instructions to check saved results or train again.
What the main files contain
| File or folder | Contents |
|---|---|
data/cases_25.csv |
Case counts for the model's 25 districts, with report volume, number and clinical year |
data/cases_26.csv |
Counts before combining the Ampara and Kalmunai reporting divisions |
data/population_audit.json |
The fixed Census 2012 population values and checks |
data/cases_25_printed_repeat_sensitivity.csv |
A different way of treating repeated reports, saved for later study |
evidence/evaluation_outputs.zip |
Original saved weights, predictions, curves, splits, setting choices, notebook and file hashes |
provenance/source_manifest.json |
Original file URLs and SHA256 fingerprints |
provenance/target_ledger.json |
How report rows were read, checked, repaired or left unresolved |
release_manifest.json |
File list and hashes for this release |
Understand the rows and missing values
The case table has 686 report positions, called report slots. There are 663 complete district rows and 23 missing rows. Empty entries mean missing data, not zero cases. We combine Ampara and Kalmunai to obtain 25 district columns. Windows containing a missing input or target row are excluded rather than filled in.
The model uses case history and fixed population scaling. Weather, measured travel and district boundaries are not model inputs. The original source ends at Vol. 53 No. 33; this is not a full 2026 calendar year.
The alternative printed-repeat table was not used to train or score the reported LEG-SEIR model. Including that table is not evidence that the model works under both reporting policies.
Training and evaluation periods
Final training uses targets through 2024. Validation in 2025 chooses the saved checkpoint. The 2026 comparison has 18 eligible starting points. They overlap, so they are not independent outbreak replications.
This is a retrospective study: the 2026 sample had already influenced earlier research choices. It is not an untouched future-outbreak test. Report dates were reconstructed from source records, not from complete records of exactly when each report first became available.
Check the saved experiment
The code repository's CPU verifier checks 915 file hashes and reloads all 265 neural fits to recalculate predictions and scores. This checks stored models; it does not train new ones. Only load checkpoints from a trusted source because these project files contain Python objects.
A hash is a file fingerprint. A matching hash confirms that bytes match the recorded copy, not that every original clinical report is correct. The exact scientific input version is pinned in the code repository's artifacts.lock.json. Later edits to this description do not change those pinned experiment files.
Original source-manifest hash: 69332f61488130cdf730e08a06f4ab417742c6b57834815d812ca7903662f239.
Original scientific code hash: 3ea0794104f30993d7b0994c17e76e95435def0ad11305cfa535c91828009ed1.
The public-table training notebook uses a new downloader but the same model and training cells. It has a different code/input identity because it starts from prepared tables rather than reading original PDFs.
Original sources
The release does not include bulk original report PDFs, weather/geographic archives, third-party paper PDFs or lecturer files. The team keeps its original source copy separately. The repository has a source downloader that checks original URLs and hashes. It reports a changed or missing file instead of silently replacing the recorded source.
Rights and intended use
MIT covers the team's original software and the team-authored experiment code/generated evidence with its accompanying notice. It does not automatically cover government reports, population tables, third-party data or the manuscript. Keep provider attribution and check each source's terms before reuse. No single unrestricted open-data license is claimed for the whole release; the card therefore uses license: other.
These files support research and result checking. They are not a clinical decision tool.
- Downloads last month
- 89