A Knot Selection Algorithm for Splines in Logistic Regression

Abstract

In ordinary logistic regression, the logit of the conditional probability of the response given the covariates is modelled as a linear function of the covariates. In this study, a more general logistic regression model is considered, where linearity is not assumed. Since the linear function of the covariates is replaced by a general function of the covariates, spline approximation is used. A knot selection algorithm is proposed to determine the knot locations in spline approximation. Simulation experiments have been carried out to check the performance of the proposed algorithm. The proposed algorithm performs reasonably well.