OmniGA is a framework for the optimization of omnivariate decision trees (ODTs) based on a parallel genetic algorithm (GA). OmniGA finds and optimizes ODT structures for binary classification problems. Reported results are obtained by using a 3-fold cross-validation technique.

The framework receives as an input a comma separated file (.csv) containing the samples defined by a set of attributes.
Rows and columns in the input csv file correspond to sample and features, respectively.
Input files must contain a header specifying a feature id for each column.
An example of an input file may be found here.

Classification problem

Upload csv file containing classification problem

Genetic algorithm parameters

Number of iterations
Population size
Crossover probability
Mutation probability
Objective function

OmniGA parameters

Add on: early stopping
Add on: deep learning
Add on: SMOTE