karoo_gp2

This is a Genetic Programming (GP) Algorithm loosely based on Karoo GP.

It currently supports regression data, and can be modified to accomodate binary and multiclass classification.

Basic Usage

import pandas as pd
from karoo_gp import Terminals, Operators, Model

# Load train data
train_data = pd.read_csv('datasets/iris.csv')
train_labels = train_data.pop('s')

# Setup model
operators = Operators.arithmetic()
terminals = Terminals(train_data.keys(), constants=[.1, .2, .3, .4, .5])
model = Model(operators, terminals)

# Train model
model.train(train_data, train_labels)
model.fittest()
>>> <Tree: '0.2*pl**2/sw ...' fitness: 37.33>

Get Started

git clone https://github.com/granawkins/karoo_gp2.git

cd karoo_gp2

Demo

For a detailed guide to the different classes and model options, refer to the Jupyter Notebook, Demo.ipynb.