Skip to content

Dataset · thuml

Time-series analysis and forecasting benchmark

thuml/Time-Series-Library

Time-series collection from the authors of Time-Series-Library for training and evaluating models across five core tasks.

What's inside

Contains tabular and text data in English, with a volume ranging from 1 to 10 million records. The dataset is part of the well-known open-source Time-Series-Library (TSLib) and is designed for developing and testing deep learning models. The collection covers five main areas: long- and short-term forecasting, imputation, anomaly detection, and classification.

What you can do with it

  • Training and comparing models for long- and short-term time-series forecasting
  • Testing anomaly detection and missing data imputation algorithms
  • Benchmarking custom architectures for sequence classification
Tasks
Time series forecasting
Languages
English
Size
1M–10M rows
Downloads this month
28.7K
Likes
14

Models trained on it

How to load it

from datasets import load_dataset

ds = load_dataset("thuml/Time-Series-Library")