House Price Prediction using Multiple Linear Regression and KNN

Abstract

The transition of BPHTB management from central taxes to regional taxes is a continuation of the regional autonomy policy. The difference between the market value and the prevailing NJOP poses a challenge for the Sintang District Government in determining the Tax Object Acquisition Value (NPOP) as the basis for imposing BPHTB. Machine learning has been extensively explored for predictions and can be an alternative that can help predict NPOP, especially house prices. This study uses backward elimination and forward selection methods to select the features used in this study and multiple linear regression and K-Nearest Neighbor methods to make house price prediction models. The results of model performance measurement using RMSE, Multiple Linear Regression method with feature selection using backward elimination resulted in a better model with an RMSE value of 44.02 (million rupiahs) and an R2 value of 0.707.