← back
COS 710 · 2026

GP Load Forecaster

Genetic Programming system for symbolic regression on UK residential electricity load data. Tournament selection, parsimony control, and tree evolution from scratch.

PythonNumPyPandasMatplotlib
shipped
gh ↗

A from-scratch GP implementation for COS 710. Features tournament selection with replacement, parsimony pressure to control tree bloat, and a full validation/test pipeline. No GP libraries — everything from tree representation to crossover and mutation is hand-rolled in Python.