CoolFace
Datasetpublic

THULab/gifteval

GIFT-Eval - TsFile Conversion This repository contains a TsFile conversion of Salesforce/GiftEval. GIFT-Eval is a benchmark for zero-shot time-series forecasting. The original dataset contains 23 datasets, about 144,000 time series and 177 million data points across seven domains and ten frequencies. Conversion Layout Each source Arrow file is converted separately and the original dataset/frequency hierarchy is preserved: data/ <dataset>.tsfile… See the full description on the dataset page: https://huggingface.co/datasets/THULab/gifteval.

sourceHugging Faceapache-2.0updated 1mo agoView on Hugging Face
0likes1.5kdownloads
Dataset Card

GIFT-Eval - TsFile Conversion

This repository contains a TsFile conversion of `Salesforce/GiftEval`.

GIFT-Eval is a benchmark for zero-shot time-series forecasting. The original dataset contains 23 datasets, about 144,000 time series and 177 million data points across seven domains and ten frequencies.

Conversion Layout

Each source Arrow file is converted separately and the original dataset/frequency hierarchy is preserved:

text
data/
  <dataset>.tsfile
  <dataset>/<freq>.tsfile

TsFile Mapping

  • item_id is a TAG.
  • start and freq are used to synthesize millisecond Time values.
  • Univariate target is stored as field target.
  • Multivariate target is stored as fields target_0, target_1, ...
  • past_feat_dynamic_real, when present, is stored as past_feat_dynamic_real_0, past_feat_dynamic_real_1, ...

start and freq are not repeated as fields because they are encoded by the generated time axis and output hierarchy.

Citation

bibtex
@article{aksu2024giftevalbenchmarkgeneraltime,
  title={GIFT-Eval: A Benchmark For General Time Series Forecasting Model Evaluation},
  author={Taha Aksu and Gerald Woo and Juncheng Liu and Xu Liu and Chenghao Liu and Silvio Savarese and Caiming Xiong and Doyen Sahoo},
  journal={arxiv preprint arxiv:2410.10393},
  year={2024}
}