DNA microarray technology is a high throughput and parallel technique for genomic investigation due to its advantages of simul-taneously surveying features of large scales complex data in biology. This paper aims to find feature subset to build the classifier for gene expression data analysis. At first, K-means clustering algorithm was carried out on the dataset of yeast cell cycle. Based on Rand cal-culation, a statistical method was used to pick out the data points (genes) for classifier design. Meanwhile, the principal component anal-ysis was applied to help to construct the classifier. For the validation of classifier built and prediction of a target subset of genes, discriminant analysis in terms of partial least square regression and artificial neural network were also performed.