Beyond Linearity: Harnessing Spline Regression Models to Capture Non-linear Relationships

Abstract

This article explores the effectiveness of spline regression model in capturing non-linear relationships in data. A comparison of spline regression with other techniques, such as linear regression, polynomial regression, generalized additive, and log-transformed models, is conducted using simulated data. The performance metrics, including AIC, BIC, RMSE, MSE, MAE, and R-squared, are used to assess the goodness of fit for each model. The results indicate that the spline regression model outperforms other methods in accurately capturing non-linear relationships. The flexibility and smoothness provided by spline regression, through the incorporation of knots, result in better-fitted lines that closely match the data. This study recommends the use of spline regression for handling non-linear data and highlights its robustness and accuracy.