Hugging Face's logo

Datasets:
smilegeng
/
fev_datasets

Tasks:
Time Series Forecasting
Modalities:
Time-series
Sub-tasks:
univariate-time-series-forecasting
multivariate-time-series-forecasting
Size:
100K<n<1M
ArXiv:
Tags:
forecasting
tsfile
format:tsfile
timeseries
License:
Dataset card Files
xet
 Community
Dataset Viewer
The dataset viewer is not available for this subset.
Cannot get the split names for the config 'default' of the dataset.
Exception:    SplitsNotFoundError
Message:      The split names could not be parsed from the dataset config.
Traceback:    Traceback (most recent call last):
                File "tsfile/tsfile_py_cpp.pyx", line 567, in tsfile.tsfile_py_cpp.tsfile_reader_new_c
              tsfile.exceptions.FileOpenError: 28: 
              
              The above exception was the direct cause of the following exception:
              
              Traceback (most recent call last):
                File "/usr/local/lib/python3.14/site-packages/datasets/inspect.py", line 286, in get_dataset_config_info
                  for split_generator in builder._split_generators(
                                         ~~~~~~~~~~~~~~~~~~~~~~~~~^
                      StreamingDownloadManager(base_path=builder.base_path, download_config=download_config)
                      ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                  )
                  ^
                File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/tsfile/tsfile.py", line 271, in _split_generators
                  scan = self._scan_metadata(all_files)
                File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/tsfile/tsfile.py", line 318, in _scan_metadata
                  with self._open_reader(file) as reader:
                       ~~~~~~~~~~~~~~~~~^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/tsfile/tsfile.py", line 742, in _open_reader
                  return TsFileReader(file)
                File "tsfile/tsfile_reader.pyx", line 323, in tsfile.tsfile_reader.TsFileReaderPy.__init__
              SystemError: <class '_weakrefset.WeakSet'> returned a result with an exception set
              
              The above exception was the direct cause of the following exception:
              
              Traceback (most recent call last):
                File "/src/services/worker/src/worker/job_runners/config/split_names.py", line 68, in compute_split_names_from_streaming_response
                  for split in get_dataset_split_names(
                               ~~~~~~~~~~~~~~~~~~~~~~~^
                      path=dataset,
                      ^^^^^^^^^^^^^
                      config_name=config,
                      ^^^^^^^^^^^^^^^^^^^
                      token=hf_token,
                      ^^^^^^^^^^^^^^^
                  )
                  ^
                File "/usr/local/lib/python3.14/site-packages/datasets/inspect.py", line 340, in get_dataset_split_names
                  info = get_dataset_config_info(
                      path,
                  ...<6 lines>...
                      **config_kwargs,
                  )
                File "/usr/local/lib/python3.14/site-packages/datasets/inspect.py", line 291, in get_dataset_config_info
                  raise SplitsNotFoundError("The split names could not be parsed from the dataset config.") from err
              datasets.inspect.SplitsNotFoundError: The split names could not be parsed from the dataset config.

Need help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.

FEV forecasting dataset collection — TsFile format

This repository is a conversion of autogluon/fev_datasets to Apache TsFile format, with 49 subsets. Each subset lives in its own directory, containing a .tsfile data file (large tables are automatically sharded into multiple .tsfile files) and a descriptive README.md.

This data was converted to a unified format by an external source and then to TsFile. License and attribution follow the original source; we claim no rights over the original data. Unless otherwise noted, the data is for research use only.

Conversion Notes

  • id (each series) → TsFile device (TAG dimension).
  • Static covariate columns → also TAG (device metadata).
  • target / dynamic covariates → measurement (FIELD).
  • timestamp → Time (INT64 milliseconds); dtypes adapt to the source (float32→FLOAT, etc.).
  • Paths mirror the original repo: <subset>/<frequency>/<frequency>.tsfile (no frequency → <subset>/<subset>.tsfile).

Subset index

Subset Frequency Series Observations Source Citation
ETT 15T, 1D, 1H, 1W 2 975,520 / 10,136 / 243,880 / 1,442 link [1]
LOOP_SEATTLE 1D, 1H, 5T 323 117,895 / 2,829,480 / 33,953,760 link [2]
M_DENSE 1D, 1H 30 21,900 / 525,600 link [2]
SZ_TAXI 15T, 1H 156 464,256 / 116,064 link [2]
australian_tourism — 89 3,204 link [3]
bizitobs_l2c 1H, 5T 1 18,648 / 223,776 link [4]
boomlet 1062, 1209, 1225, 1230, 1282, 1487, 1631, 1676, 1855, 1975, 2187, 285, 619, 772, 963 1 344,064 / 868,352 / 802,816 / 376,832 / 573,440 / 884,736 / 418,520 / 1,046,300 / 272,012 / 392,325 / 523,100 / 1,228,800 / 851,968 / 1,097,728 / 458,752 link [5]
ecdc_ili — 25 4,797 link —
entsoe 15T, 1H, 30T 6 6,310,512 / 1,577,592 / 3,155,220 link [6]
epf_be — 1 157,248 link [7]
epf_de — 1 157,248 link [7]
epf_fr — 1 157,248 link [7]
epf_np — 1 157,248 link [7]
epf_pjm — 1 157,248 link [7]
ercot 1D, 1H, 1M, 1W 8 51,616 / 1,238,976 / 1,688 / 7,368 link —
favorita_stores 1D, 1M, 1W 1,579 10,661,408 / 255,798 / 1,136,880 link [8]
favorita_transactions 1D, 1M, 1W 51 258,264 / 5,508 / 24,480 link [8]
fred_md_2025 — 1 100,548 link [9]
fred_qd_2025 — 1 65,170 link [10]
gvar — 33 52,866 link [11]
hermes — 10,000 5,220,000 link [12]
hierarchical_sales 1D, 1W 118 215,350 / 30,680 link [4]
hospital — 767 64,428 link [4]
hospital_admissions 1D, 1W 8 13,846 / 1,968 link [13]
jena_weather 10T, 1D, 1H 1 1,106,784 / 7,686 / 184,464 link [4]
kdd_cup_2022 10T, 1D, 30T 134 47,273,860 / 325,620 / 15,755,720 link [14]
m5 1D, 1M, 1W 30,490 428,849,460 / 13,805,685 / 60,857,703 link [15]
proenfo_bull — 41 2,877,216 link [16]
proenfo_cockatoo — 1 105,264 link [16]
proenfo_gfc12 — 11 867,108 link [16]
proenfo_gfc14 — 1 35,040 link [16]
proenfo_gfc17 — 8 280,704 link [16]
proenfo_hog — 24 2,526,336 link [16]
proenfo_pdb — 1 35,040 link [16]
redset 15T, 1H, 5T 126 1,052,371 / 283,070 / 2,960,408 link [17]
restaurant — 817 294,568 link [18]
rohlik_orders 1D, 1W 7 115,650 / 15,316 link [19]
rohlik_sales 1D, 1W 5,390 74,413,935 / 10,516,770 link [20]
rossmann 1D, 1W 1,115 7,352,310 / 889,770 link [21]
solar 1D, 1W 137 50,005 / 7,124 link [4]
solar_with_weather 15T, 1H 1 1,986,000 / 496,480 link —
uci_air_quality 1D, 1H 1 5,057 / 121,641 link [22]
uk_covid_nation 1D, 1W 4 41,216 / 5,936 link —
uk_covid_utla 1D, 1W 214 308,786 / 44,448 link —
us_consumption 1M, 1Q, 1Y 31 24,552 / 8,122 / 1,984 link [23]
walmart — 2,936 4,609,143 link [24]
world_co2_emissions — 191 11,460 link —
world_life_expectancy — 237 17,538 link [25]
world_tourism — 178 3,738 link [26]

Reading example 

from tsfile import TsFileReader

reader = TsFileReader("<freq>.tsfile")
schemas = reader.get_all_table_schemas()
# Table name: <see each subset README>; columns: see "Column meanings" in the subset README.

Citation

The original collection fev-bench:

@article{shchur2025fev,
  title={{fev-bench}: A Realistic Benchmark for Time Series Forecasting},
  author={Shchur, Oleksandr and Ansari, Abdul Fatir and Turkmen, Caner and Stella, Lorenzo and Erickson, Nick and Guerron, Pablo and Bohlke-Schneider, Michael and Wang, Yuyang},
  year={2025},
  eprint={2509.26468},
  archivePrefix={arXiv},
  primaryClass={cs.LG}
}
Downloads last month
260
Total file size:
758 MB

Papers for smilegeng/fev_datasets