Computing importance value of medical data parameters in classification tasks and its evaluation using machine learning methods

Abstract

This paper aims to evaluate the importance values of medical data parameters for further classification tasks. One of the steps of proposed methodology for analyzing medical data is initial data analysis. One part of the initial data analysis is to determine the importance rate of parameters in given data set. The reason behind this step is to provide overview of the parameters and the idea of choosing right predictors for classification task. Statistica 13 software provides a tool for determining the importance rate of each data parameter, which can be found in feature selection module. However, it is not always clear whether is the importance rate correct or not.