Nixtla/statsforecast

Health v2 · maintenance only

Adoption not part of the score

4887 stars · 385 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-29, confidence not recorded

StatsForecast is a Python library providing fast, scalable implementations of statistical and econometric univariate time series forecasting models such as AutoARIMA, AutoETS, AutoCES, MSTL, and Theta. It supports probabilistic forecasting, anomaly detection, exogenous variables, and distributed execution on Spark, Dask, and Ray with a familiar sklearn-like fit/predict API.

Use cases

• forecast sales or demand across millions of time series
• fit automatic ARIMA models faster than pmdarima or statsmodels
• generate prediction intervals and probabilistic forecasts
• detect anomalies in time series data
• benchmark forecasting models like naive, seasonal naive, and ETS
• replace Prophet with a faster statistical alternative
• run distributed forecasting on Spark, Dask, or Ray clusters
• perform time series cross-validation and model selection

When to choose

• you need fast, accurate univariate statistical forecasting at scale
• you want sklearn-style fit/predict syntax for many time series
• you need prediction intervals, conformal prediction, or anomaly detection
• you want to benchmark against naive and seasonal baselines
• you need distributed forecasting across large datasets

When to avoid

• you need deep learning or neural forecasting models (use NeuralForecast instead)
• you need multivariate forecasting with cross-series dependencies
• you work outside the Python ecosystem
• you need a hosted foundation-model forecasting API (consider TimeGPT)

Facets

library · maturity stable

machine-learning data-science benchmarking time-series data-science machine-learning python forecasting arima ets exponential-smoothing theta mstl probabilistic-forecasting anomaly-detection distributed-computing spark dask ray sklearn-api

Member repositories

For agents

markdown · JSON · MCP: product_card(name="Nixtla/statsforecast")