Statistical class prediction method for efficient microarray gene expression data sample classification

Cancer classification is routinely done using gene expression data. With microarray technology, monitoring thousands of genes is an easy task. The reliable and precise classification of different tumour types is very important in cancer classification and drug discovery which is useful in providing better treatment. Microarray gene expression data analysis is extensively used for human cancer diagnosis and classification. Various methods of classification from the field of statistics and machine learning have been used to classify cancer microarray data. However, the large number of features with very few samples in the data is a challenge to the existing classification methods. In this work, our experiments are based on the statistical method of confidence interval for predicting sample class labels. Experiments show that the proposed method will achieve high classification accuracies with very few genes.