Spline regression represents a powerful nonparametric approach to modeling complex relationships between predictor and response variables. This comprehensive review examines the theoretical foundations, practical implementations, and diverse applications of spline-based regression techniques. We explore various spline types including polynomial splines, Bsplines, natural splines, and smoothing splines, discussing their mathematical properties, computational aspects, and relative advantages. Through detailed analysis of knot selection strategies, regularization techniques, and model selection criteria, we demonstrate how splines provide flexible alternatives to traditional parametric regression models. Case studies across multiple domains illustrate the practical utility of spline regression in capturing non-linear patterns while maintaining interpretability and computational efficiency. This review serves as a comprehensive resource for researchers and practitioners seeking to understand and implement spline-based regression methodologies.
Rayarao et al. (Thu,) studied this question.