Abstract: This paper applies the Gram-Schmidt process into the logistical regression model and proposes the Schmidt-logistical regression method, which chooses the most explainable information from the original independent variable set, omits useless or redundant information, and transforms the selected independent variables into orthogonal ones. In this way, this method can not only realize variable selection in logistic regression, but also obtains a clear-structured model, which can be easily explained. This method is applied in the analysis of style investing, effectively performing variables selection and judging growth/value property of stocks.