CausalPy
CausalPy copied to clipboard
pymc-labs
A Python package for causal inference in quasi-experimental settings
* Closes #562 * Adds `RESEARCH.md` (First-pass attempt. 100% sure changes will be required. But a decent starting point.) * Adds reference to that in `CONTRIBUTING.md` ---- 📚 Documentation preview...
## Summary Extend the `InterruptedTimeSeries` class to support a three-period design: **pre-intervention**, **intervention**, and **post-intervention** periods. This enables analysis of temporary interventions and measurement of long-term effect persistence after interventions...
### Summary Add a developer-oriented `RESEARCH.md` file at the repository root that documents the internal structure of CausalPy: its main modules, core abstractions, public API surface, typical workflows, and key...
AFTER https://github.com/pymc-labs/CausalPy/issues/52 IS CLOSED At that point we can add additional time series decomposition plots, such as trend, seasonality (monthly/yearly), residuals. This could be expanded beyond just the time series...
A few errors have crept in here. We've got a demo of custom priors in here, attempting to model the data in the original scale. But the model formula has...
Variable selection priors were shown to be useful for causal inference here: https://github.com/pymc-labs/CausalPy/pull/537 and I can think of more use cases especially for synthetic control methods. I think the functionality...
## New Feature: `InterventionTimeEstimator` for Unknown Treatment Timing This PR introduces a new model, `InterventionTimeEstimator`, designed to estimate when an intervention has an effect in a time series — especially...
### Summary Introduce a small set of Claude Skills to the CausalPy repository so that developers and users can ask Claude to perform common CausalPy-specific tasks in a consistent, project-aware...
## Problem Seems like `causalpy/data/simulate_data.py` module has some reproducibility issues and a bit of refactoring that could be done ### Reproducibility The module declares a seeded RNG but doesn't use...
At the moment we have no in-built plotting capabilities to plot the model _prior_ predictions. Adding this would be very useful. The most obvious thing we could do is to...