A Comparative Study of Classification Algorithms for Cancer Datasets

Abstract

Classification algorithms of data mining have been successfully applied in the recent years to predict cancer based on Micro-array Gene Expression Data. Various classification algorithms can be applied on such Micro-array Gene Expression Data to devise methods that can predict the occurrence of cancer. In this study, Comparison of five different algorithms i.e. Voted Perceptron, LWL, DECORATE, Random Forest and RIDOR is presented. The main aim of this study is to evaluate the performance of those five algorithms for different cancer datasets with different dimensions. The datasets used for the study are chosen such a way that they differ in size, mainly in the terms of number of instances and number of attributes. When comparing the performance of all five algorithms, Random Forest is found to be the better algorithm in most of the cases.