Perbandingan Model Regresi Linier dan Nonlinier untuk Prediksi Suhu Udara di Provinsi Lampung

Abstract

This study aims to analyze the relationship between air temperature and variables such as air pressure, humidity, and wind speed using both linear and nonlinear regression approaches. Three models are compared: multiple linear regression, second-degree polynomial regression, and spline regression. The analysis results show that the spline regression model delivers the best performance, with an RMSE of 0.556 and an MAE of 0.427, which are lower than those of the polynomial regression (RMSE = 0.746; MAE = 0.570) and multiple linear regression (RMSE = 0.972; MAE = 0.715). Overall, spline regression effectively captures the complex nonlinear patterns in the data, including the presence of turning points in the influence of predictor variables on temperature. These findings indicate that nonlinear approaches, particularly spline regression, are more accurate and appropriate for modeling climate phenomena characterized by nonlinear relationships among variables.