-
Notifications
You must be signed in to change notification settings - Fork 98
Formulaic and statsmodels dependencies #369
Copy link
Copy link
Labels
Effort > Moderate 🐕Mid-sized tasks estimated to take a few days to a few weeks.Mid-sized tasks estimated to take a few days to a few weeks.Impact > Moderate 🔶User-visible but non-breaking change. Treated like a minor version bump (e.g., 0.6.5 → 0.7.0).User-visible but non-breaking change. Treated like a minor version bump (e.g., 0.6.5 → 0.7.0).Refactor 🔧Code level improvements that restructure existing logic without changing behavior.Code level improvements that restructure existing logic without changing behavior.
Milestone
Description
Metadata
Metadata
Assignees
Labels
Effort > Moderate 🐕Mid-sized tasks estimated to take a few days to a few weeks.Mid-sized tasks estimated to take a few days to a few weeks.Impact > Moderate 🔶User-visible but non-breaking change. Treated like a minor version bump (e.g., 0.6.5 → 0.7.0).User-visible but non-breaking change. Treated like a minor version bump (e.g., 0.6.5 → 0.7.0).Refactor 🔧Code level improvements that restructure existing logic without changing behavior.Code level improvements that restructure existing logic without changing behavior.
chainladder currently relies on
patsyfor Wilkinson formulas (R-style formulas). This gets used inBarnettZehnwirthandTweedieGLM. However,patsyis no longer maintained and is instead replaced by formulaic. This has expanded features including pickling of estimators that use these formulas.TweedieGLM could be greatly simplified (and relied on in the
BootstrapODPSampleif we had a few additional statistics associated with the model fit.sklearn.linear_model.TweedieRegressoronly produces coefficients. statsmodels has a full GLM suite that would elminate the need for us to calculate our own statistics such as the pearson residuals or hat matrix.All
statsmodelsdependencies arechainladderdependencies so this does not cause an alarming increase in dependency footprint.