r2py

An AI-assisted, phase-by-phase conversion of R's rpart (CART decision trees) to a standalone Python package, r2py_rpart, reusing rpart's original C source via a fake R-API header layer and cffi, with a full technical report documenting the methodology.

An AI-assisted port of R's KernSmooth package to Python, with the full dependency-analysis, conversion, and validation pipeline used to produce it — compiles the original Fortran routines via f2py and reimplements the R wrapper layer in typed, tested Python.

Python port of R's KernSmooth: binned kernel density estimation, local polynomial regression, and plug-in bandwidth selection, wrapping the original Fortran routines via f2py.

Python port of R's rpart: recursive partitioning and regression trees, with a compiled C core (via cffi) and R-parity Python functions for fitting, pruning, predicting, and plotting trees. Supports classification, regression, Poisson/survival, and user-defined splits. GPL-2.0-or-later, derived from Therneau & Atkinson's rpart.