Human Ovarian carcinoma microarray data analysis based on Support Vector Machines with different kernel functions

According to the statistics from the Department of Health, occurrence rate and mortality rate of ovarian cancer are both ranked in the top ten in Taiwan. Moreover, the mortality rate of ovarian cancer is at the first place among gynecologic cancer. In this research, we use ovarian cancer gene chip as the base of database analysis in order to solve the problems such as large number of gene chip variables, insufficient number of samples. This research use a gene chip database which contains 9,600 ovarian DNA expressions from 41 samples. Then we use linear regression and variance analysis (ANOVA) for data pre-processing for the purpose of lowering the number of genes and find the most valuable genes. Finally the information database is examined and classified by Support Vector Machine (SVM), and conduct the comparison of different results of Kernel Function. Our research discovers that the SVM has considerably fine effect in classification and when different Kernel Function appears, the results will change too. At last we have discussion of the final result for identifying the most precise and efficient Kernel Function