Figure 1: Statistical power of the training sets for the stepwise method and each penalized method using uncensored data sets. Numbers of candidate genes p = 100 and 1000, and true values of the coefficients = 0.2 and 0.4.