A flexible, intuitive and fast forecasting library.


Easy to start

Forecasting takes only a few lines of code. Start by picking an algorithm and parameter template to run through our forecast pipeline.


Fast algorithm

The flagship algorithm, Silverkite, runs fast to facilitate interactive prototyping, grid search, and benchmarking.


Flexible design

Silverkite provides regressors for trend, seasonality, and much more. You may add your own features and fit the machine learning model of your choice.


Interpretable approach

For trusted decision-making and tuning. Changepoint detection and exploratory plots build intuition; component plots and R-like model summaries explain results.