Home
Tests one factor. Screens a thousand.
Does this factor have predictive value?
factrix is a Polars-native toolkit for factor inference and batch screening. It provides methods for cross-sectional, event, and common factors while keeping estimates, p-values, diagnostics, and warnings explicit.
Start here¶
| If you want | Go to |
|---|---|
| Install and run a complete example | Installation · Quickstart |
| Prepare and validate a panel | Preparing data |
| Match a research question to a metric | Choosing a metric |
| Understand factor types and dispatch | Concepts · Panel vs timeseries |
| Read estimates, metadata, and warnings | Reading results |
| Screen many factors with FDR control | Multi-factor screening |
| Look up a metric or public symbol | Metric applicability · API reference |
| Decide whether factrix fits your workflow | Where factrix fits |
Workflow¶
- Prepare a long Polars frame on a consistent period grid.
- Choose metrics that match the factor scope, density, and data structure.
- Run
evaluate()and inspect each metric's estimate, sample metadata, and warnings. - When screening a family of candidates, apply the multi-factor FDR helpers.
The API reference lists every entry point. Contributors should start with the development guide and architecture.