Microarray Data Analysis and Model Construction Based on Oversampling Approach and Decision Tree

Primary skin cancer can be divided into basal cell carcinoma(BCC), squamous cell carcinoma and melanoma three types. Especially, BCC incidence is as high as 80% with 10% global incidence increasing each year. Although BCC has low metastatic and lower mortality than the other two types, this malignant tumor has a very high incidence. Therefore places a great burden on global healthcare expenditures. Imiquimod and Resiquimod are therapies for BCC. Imiquimod has been shown to be highly effective in the treatment of localized BCC. Therefore, the purpose of this study is to analyze the microarray of therapeutic drugs for BCC using machine learning algorithm and constructing prediction model. In this study, microarray raw data with giving Imiquimod and Resiquimod to BCC, which contained 30,968 sets of gene expression. There are two parts of this study: First, the three gene selection algorithms were employed, such as linear regression analysis, information gain and gain ratio, to screen out the biomaker. Second, we use decision tree algorithms: C4.5 to construct the prediction model and discuss classification accuracy.