functime is a library and toolkit for machine learning on time series, designed to work at scale—from quick setup to evaluating thousands of forecasts in parallel.
End-to-end time-series forecasting
You can get started with a simple install (`pip install functime`) and follow the documentation and tutorial that walks through a complete forecasting workflow.
- Data preparation for time-series modeling
- Training and generating forecasts
- Comparing results across models
Evaluate and compare forecasts in bulk
functime includes utilities for scoring, ranking, and visualization so you can assess model quality across large collections of time series at once.
- Forecast scoring and evaluation procedures
- Ranking and comparison across many series
- Visual tools to review performance
LLM Forecast Analysts
The “LLM Forecast Analysts” module acts as an AI copilot for interpreting forecasts and diagnostics.
- Trend and seasonality analysis
- Identifying possible causal factors
- Comparing different forecasts
- API reference and a GitHub repository for developers
