Skip to content

Python Time Series Libraries

Curated list of Python libraries for time series analysis and forecasting.

Statistical Methods

  1. statsmodels - Classical statistical models: GitHub | Docs ARIMA, SARIMAX, VAR, state space models, statistical tests
  2. pmdarima - Auto-ARIMA for Python: GitHub | Docs Automated ARIMA modeling, seasonal decomposition, scikit-learn compatible
  3. Prophet - Facebook's forecasting tool: GitHub | Docs Business forecasting with seasonality, holidays, changepoints

Machine Learning

  1. sktime - Unified ML framework GitHub | Docs Classification, regression, forecasting, scikit-learn compatible

  2. tslearn - Time series ML GitHub | Docs Clustering, classification, DTW, shapelet learning

  3. GluonTS - Probabilistic forecasting GitHub | Docs DeepAR, Transformer models, MXNet/PyTorch backends

  4. aeon - ML toolkit (sktime successor) GitHub | Docs Classification, regression, clustering, next-gen sktime

  5. pyts - Time series classification GitHub | Docs Imaging, shapelet transform, bag-of-words, classification

  6. PyPOTS - Partially observed time series: GitHub | Docs Imputation, forecasting, classification for incomplete data

  7. deeptime - Dynamical systems GitHub | Docs Markov models, Koopman theory, molecular dynamics

Deep Learning

  1. PyTorch Forecasting - Neural forecasting: GitHub | Docs N-BEATS, TFT, DeepAR, NHiTS implementations
  2. Darts - User-friendly forecasting GitHub | Docs Classical + neural methods, backtesting, ensemble models
  3. NeuralProphet - Neural Prophet GitHub | Docs Prophet reimplemented with PyTorch, autoregression support
  4. Time-Series-Library - SOTA models collection: GitHub Comprehensive library: iTransformer, TimesNet, PatchTST, DLinear, etc.
  5. NeuralForecast - Nixtla's neural methods: GitHub | Docs RNN, LSTM, N-BEATS, N-HiTS, TFT implementations
  6. Flow Forecast - Deep learning forecasting: GitHub | Docs Multi-variate forecasting, transformers, deployment ready
  7. InceptionTime - Deep learning classifier GitHub SOTA time series classification with Inception architecture
  8. TimeMixer - Mixing forecasting model: GitHub Multi-scale mixing for long-term forecasting

Foundation Models

  1. Time-LLM - LLM for time series: GitHub Large language models adapted for time series forecasting

  2. Large-Time-Series-Model - Foundation model GitHub Pre-trained model for zero-shot forecasting

Feature Engineering

  1. tsfresh - Automatic feature extraction GitHub | Docs Extract 700+ time series features automatically
  2. tsfeatures - Feature extraction GitHub | PyPI Fast feature computation for time series
  3. Catch22 - 22 canonical features GitHub | Docs Minimal feature set for classification

Anomaly Detection

  1. stumpy - Matrix profile GitHub | Docs Pattern discovery, anomaly detection, motif search
  2. PyOD - Outlier detection GitHub | Docs 40+ anomaly detection algorithms

Utilities

  1. pandas - Time series data structures GitHub | Docs DatetimeIndex, resampling, rolling windows

  2. scipy - Signal processing GitHub | Docs Filters, spectral analysis, interpolation

  3. matplotlib - Time series plotting GitHub | Docs Visualization and plotting

  4. Pathway - Real-time data processing GitHub | Docs Streaming time series, real-time ML pipelines

Other tools are added regularly ...