easy-geppy 0.1.2

EasyGeppy is an easy to use programming interface for Geppy

EasyGeppy

EasyGeppy is an easy to use programming interface for Geppy from Shuhua Gao [1], and proposed by C. Ferreira in 2001 [2] for Gene Expression Programming (GEP).

EasyGeppy provides a minimized and pre-defined pipeline setup for solving simple and multiple regression problems using Geppy along with Pandas [3] and Numpy [4].

The pipeline is based on the following Shuhua Gao's notebook: Simple mathematical expression inference.

Nonetheless, EasyGeppy allows you to set your custom configuration to its setup by accessing the class EasyGeppy attributes.

Feel free to contribute.

How to install

pip install easy_geppy

How to use

#import
from easy_geppy import EasyGeppy

# Initialize
egp = EasyGeppy(df, #Pandas DataFrame
                 x_columns=['column1','column2','column3'],
                 y_column='column_y')

egp.default_initialization()

# Train
egp.launch_evolution(n_pop=300, n_gen=100)

# Get resulting function for making predictions
best_func = egp.get_best_solution_as_function()

# Make predictions
df['y_predicted'] = best_func(df)

# Get symbolic representation of the resulting function
egp.get_best_solution_simplified()

Example

1. EasyGeppy-Example

Reference

[1] Shuhua Gao (2020) Geppy [Source code]. https://github.com/ShuhuaGao/geppy. [2] Ferreira, C. (2001). Gene Expression Programming: a New Adaptive Algorithm for Solving Problems. Complex Systems, 13. [3] McKinney, W. & others, 2010. Data structures for statistical computing in python. In Proceedings of the 9th Python in Science Conference. pp. 51–56. [4] Harris, C.R., Millman, K.J., van der Walt, S.J. et al. Array programming with NumPy. Nature 585, 357–362 (2020). DOI: 10.1038/s41586-020-2649-2. (Publisher link).