Iterative L1/2 Regularization Algorithm for Variable Selection in the Cox Proportional Hazards Model

摘要

In this paper, we investigate to use the L_(1/2) regularization method for variable selection based on the Cox's proportional hazards model. The L_(1/2) regularization method is a reweighed iterative algorithm with the adaptively weighted L_1 penalty on regression coefficients. The algorithm of the L_(1/2) regularization method can be easily obtained by a series of L_1 penalties. Simulation results based on standard artificial data show that the L_(1/2) regularization method can be more accurate for variable selection than Lasso and adaptive Lasso methods. The results from Primary Biliary Cirrhosis (PBC) dataset indicate the L_(1/2) regularization method performs competitively.