# timeseriesAI/tsai

Time series Timeseries Deep Learning Machine Learning Python Pytorch  fastai | State-of-the-art Deep Learning library  for Time Series and Sequences in Pytorch / fastai

Repository: https://github.com/timeseriesAI/tsai
Canonical: https://ross.abutalabs.com/products/tsai
Homepage: https://timeseriesai.github.io/tsai/
Language: Jupyter Notebook
License: Apache-2.0
License Family: permissive
Topics: time-series-classification, deep-learning, fastai, pytorch, timeseries, time-series, sequential, time-series-analysis, transformer, cnn, rnn, state-of-the-art, self-supervised, classification, regression, forecasting, inceptiontime, rocket, machine-learning, python
Last push: 2026-07-23T21:55:35+00:00

## Health v2 (maintenance only)
Score: 84/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 94, release rhythm 62, longevity 100
- inputs: {"age_days": 2529, "days_push": 41, "days_rel": 98, "gap_med": 149, "n_releases_24m": 4}
- flags: none
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 6111, forks 721 (observed 2026-08-28T04:09:35.995766+00:00)

## What it is
tsai is an open-source deep learning library built on PyTorch and fastai for time series and sequential data tasks such as classification, regression, forecasting, and imputation. It provides state-of-the-art models (PatchTST, InceptionTime, ROCKET, RNNs with attention), built-in benchmark datasets, sklearn-style pipeline transforms, and walk-forward cross-validation.

## Use cases
- classify time series with deep learning in pytorch
- forecast time series with PatchTST or transformer models
- train a model for time series regression
- benchmark time series classification models like InceptionTime and ROCKET
- impute missing values in time series data
- do walk-forward cross-validation for time series forecasting
- download UCR/UEA time series datasets easily

## When to choose
- you want state-of-the-art deep learning models for time series in Python
- you already use PyTorch or fastai and need time series support
- you need quick access to standard time series benchmark datasets
- you want sklearn-like pipelines and walk-forward validation for forecasting

## When to avoid
- you need classical statistical forecasting methods like ARIMA or Prophet
- you work outside Python or without GPU support
- you need a lightweight tabular ML tool rather than deep learning
- you rely on conda distribution, which is no longer updated

## Facets
- artifact type: library
- maturity: active
- function: deep-learning, machine-learning, data-science
- domain: time-series, deep-learning, machine-learning
- platform: python
- tags: pytorch, fastai, time-series-classification, time-series-regression, forecasting, imputation, transformer, patchtst, inceptiontime, rocket, self-supervised, gpu

## Member repositories
- timeseriesAI/tsai (main) score 84

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:09:35.995766+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-29T17:48:11.723396+00:00, confidence not recorded.
  - readme: https://github.com/timeseriesAI/tsai (fetched 2026-08-28T04:09:35.995766+00:00, sha 30a4c4a66185)
  - homepage: https://timeseriesai.github.io/tsai/ (fetched 2026-08-29T08:44:44.025010+00:00, sha 596227746ad6)
  - registry_pypi: https://pypi.org/pypi/tsai/json (fetched 2026-08-29T08:44:44.027855+00:00, sha b2edb563d1e6)
- Data as of 2026-08-30T08:39:29.467469+00:00.