Abstract: Microarray data has been widely applied to cancer classification, where the purpose is to classify and predict the category of a sample by its gene expression profile. The cancer classification is required to identify the significant genes that are a good subset of original features. For evaluating the goodness of a subset of features, the feature selection methods fall into two broad categories: the filter approach and the wrapper approach. Since wrapper methods are computationally very intensive, filter approach is chosen for selecting the most informative genes. DNA microarray is a gene chip which consists of expression levels for a huge number of genes a relatively small number of samples. However, only a small number of genes contribute in accurate classification of cancer. Therefore, the challenging task is to identify a small subset of informative genes which has maximum amount of information about the class. The feature selection method is used to find the informative genes which helps to minimize the classification errors. The hybrid correlation methods are used to find out the correlated and negative correlated features. The classifier Support Vector Machine along with Decision Tree Algorithm is proposed to classify the features. The result is compared with the performance of neural network classifier which gives better accuracy in positive correlated features than hybrid correlated and negative correlated features.

Keywords: DNA Microarray, Classification, Correlation, Neural Network, Backpropagation Algorithm.