RIS is a novel instance selection algorithm that attributes a score per instance that depends on its relationship with all other instances in the training set. To read more about RIS algorithm, please consider the following paper:

CAVALCANTI, George DC; SOARES, Rodolfo JO. Ranking-based Instance Selection for Pattern Classification. Expert Systems with Applications, p. 113269, 2020.

These instructions will get you replicate the experiments carry out on your local machine and reported in RIS paper.

Firstly, go to folder src/algorithm/instance_selection/ris and run the code below:

user@src/algorithm/instance_selection/classification/ris> python setup.py build_ext --inplace

This code above will compile helper functions of RIS, a cython implementation.

All datasets used in the experiments are available in Knowledge Extraction based on Evolutionary Learning.

To apply RIS method over a dataset, you should run the follow code on your machine:

user@ris> python run.py

The code above will running RIS implementation over all datasets listed in datasets variable inside the run.py