taharnbl/fm_impute_bench

Data format and usage

TODO

Dataset statistics

dataset	freq	# series	series length	# test windows	source
BDG2-Bear	1H	91	17,544	7,522	https://huggingface.co/datasets/Salesforce/lotsa_data
BDG2-Rat	1H	280	17,544	24,915	https://huggingface.co/datasets/Salesforce/lotsa_data
Borealis	1H	15	7,447	77	https://huggingface.co/datasets/Salesforce/lotsa_data
Covid19 Energy	1H	1	31,912	195	https://huggingface.co/datasets/Salesforce/lotsa_data
GFC12 Load	1H	20	39,414	4,960	https://huggingface.co/datasets/Salesforce/lotsa_data
Hog	1H	24	17,544	2,310	https://huggingface.co/datasets/Salesforce/lotsa_data
Ideal	1H	217	16,167	156	https://huggingface.co/datasets/Salesforce/lotsa_data
PDB	1H	1	17,520	96	https://huggingface.co/datasets/Salesforce/lotsa_data
KDD Cup2022	10min	134	35,279	2,546	https://huggingface.co/datasets/Salesforce/lotsa_data
ERA5 geopotential	1H	500	8,736	19,000	https://huggingface.co/datasets/Salesforce/lotsa_data
ERA5 humidity	1H	500	8,736	19,000	https://huggingface.co/datasets/Salesforce/lotsa_data
ERA5 temperature	1H	500	8,736	19,000	https://huggingface.co/datasets/Salesforce/lotsa_data
ERA5 wind speed	1H	500	8,736	19,000	https://huggingface.co/datasets/Salesforce/lotsa_data
Oikolab weather	1H	8	100,057	5,288	https://huggingface.co/datasets/Salesforce/lotsa_data
Pedestrian counts	1H	66	96,400	7,733	https://huggingface.co/datasets/Salesforce/lotsa_data
Traffic	1H	861	17,544	82,719	https://huggingface.co/datasets/Salesforce/lotsa_data
PEMS BAY	5min	325	52,128	2,275	https://huggingface.co/datasets/Salesforce/lotsa_data
PEMS 03	5min	358	26,208	358	https://huggingface.co/datasets/Salesforce/lotsa_data
SHMETRO	15min	576	8,809	576	https://huggingface.co/datasets/Salesforce/lotsa_data
ETT1-15T	15min	7	69,680	1,050	https://huggingface.co/datasets/Salesforce/GiftEval/tree/main
ETT1-1H	1H	7	17,420	1,092	https://huggingface.co/datasets/Salesforce/GiftEval/tree/main
ETT2-15T	15min	7	69,680	1,050	https://huggingface.co/datasets/Salesforce/GiftEval/tree/main
ETT2-1H	1H	7	17,420	1,092	https://huggingface.co/datasets/Salesforce/GiftEval/tree/main
Solar-1H	1H	137	8,760	8,768	https://huggingface.co/datasets/Salesforce/GiftEval/tree/main
Jena Weather 10T	10min	21	52,704	1,428	https://huggingface.co/datasets/Salesforce/GiftEval/tree/main
Jena Weather 1H	1H	21	8,784	1,344	https://huggingface.co/datasets/Salesforce/GiftEval/tree/main
LOOP SEATTLE 5T	5min	323	105,120	21,964	https://huggingface.co/datasets/Salesforce/GiftEval/tree/main
LOOP SEATTLE 1H	1H	323	8,760	20,672	https://huggingface.co/datasets/Salesforce/GiftEval/tree/main
M_DENSE	1H	30	17,520	4,710	https://huggingface.co/datasets/Salesforce/GiftEval/tree/main
Enedis LDM Small	30min	500	17,424	20,500	https://zenodo.org/records/15232742
London Smart Meters Small	30min	500	22,000	25,779	https://huggingface.co/datasets/autogluon/chronos_datasets/tree/main/monash_london_smart_meters
Spanish Energy	1H	9	35,064	1,962	https://www.kaggle.com/datasets/nicholasjhana/energy-consumption-generation-prices-and-weather
Weather	1H	11	35,064	2,398	https://github.com/zhouhaoyi/Informer2020

Citation

If you use fm_impute_bench for research purposes, please consider citing the associated paper:

@article{lenaour2026are,
    title={Are Time-Indexed Foundation Models the Future of Time Series Imputation?},
    author={Le Naour, Etienne and Nabil, Tahar and Petralia, Adrien and Agoua, Ghislain},
    journal={Transactions on Machine Learning Research},
    issn={2835-8856},
    year={2026},
    url={https://openreview.net/forum?id=cTk56KpsP5}
}