Novel method for missing value estimation in gene expression profile based on support vector regression

Gene expression profile is a kind of useful biological resource in recent years. The original data is usually in the form of large matrix with missing values in it. Missing value estimation method is in active demand as many downstream analysis methods suffered from the missing values. Several methods dealing with this problem have been reported (proposed) recently. We proposed a novel missing value estimation method based on support vector regression. Various parameter sets and input coding schemes were employed, and the performance of our method was evaluated over different data sets by comparing with other existing methods. Results show that our method is comparative, if not better than, those previous methods