Evaluating Machine Learning Models in Housing Price Forecasting

Abstract

This study aims to compare the performance of different machine learning algorithms in predicting housing prices and to construct an experimental process based on data preprocessing, feature selection, and comparative analysis of multiple regression models. Initially, a Random Forest will filter out the ten most significant variables. This variable will be utilized to build Multivariable Linear Regression (MLR). To find the best subset, lasso, best subset, and 5-fold Cross-Validation methods will be applied to find the final model. At the same time, this study incorporates training and testing using models such as K-Nearest Neighbors (KNN), Support Vector Machine (SVM), Bagging, Random Forest (RF), Boosting, and Bayesian Additive Regression Trees (BART). By comparing the error performance of the model on the training set and the test set, the mean squared error (MSE) will be calculated to evaluate the strengths and weaknesses of each model in housing price prediction. The goal is to find the modeling method and variable combination that best suits the data set.