Cancer survivability prediction using random forest and rule induction algorithms

Abstract

The second largest cause of death in Palestine is Cancer at a rate 12.4% of all deaths. Predicting the survivability of a disease is one of the most interesting purposes of developing a medical data mining applications. This paper applies two classification models (Rule Induction and Random Forest) on the Gaza Strip 2011 cancer patient's dataset, to predict the survivability of cancer patients. The experiments were conducted on the dataset using RapidMiner tool which is used to build the classification models and to measure the performance of them in terms of time consumption and model accuracy. We found that the two algorithms have a convergent accuracy while random forest was less time consuming than rule induction. The rule induction algorithm has the accuracy of 73.63% while Random Forest scores 74.6% accuracy.