Predicting Cancer from Microarray Data Using Statistical Method

The development of microarray technologies has made it to obtain gene expression pattern of thousands of genes in a single cell simultaneous. Based on such microarray data, assessment of gene variations including classification and developmental status of cancer cells are possible. The objective of this paper is to predict and classify gene expression information by means of analysis of microarray data, using statistics method. We try to explore significant features and classifiers using Leukemia cancer dataset. In this work, information theory is used to select significant features in the preprocessing step. We then use discriminant analysis and decision tree and logistic regression to classify the selected features and compare their precision and recall. In our experiments, discriminant classifier outperformed the other classifiers.