pmdarima

What it is

A statistical Python library that fills gaps in Python's time series analysis capabilities, including an equivalent of R's auto.arima function. It provides stationarity and seasonality tests, differencing utilities, Box-Cox and Fourier transformations, seasonal decompositions, cross-validation tools, and scikit-learn-style pipelines. Users coming from a scikit-learn background will find the wrapping of statsmodels familiar. The library depends on statsmodels under the hood, so its forecasting scope is limited to statistical ARIMA-style modeling rather than broader machine learning approaches.