Comparison Of Linear and Nonlinear Models in Predicting House Prices

Nowadays, the increase in housing prices has led to a heavier financial burden on ordinary citizens, affecting their quality of life. This research used linear regression, random forest, and Support Vector Machine (SVM) to predict housing prices using the Boston housing price dataset, and compared the advantages and disadvantages of them. The paper compares the performance of linear and nonlinear models in predicting housing prices to reflect the importance and magnitude of various factors that affect housing prices. The overall research demonstrates that nonlinear models have unique advantages over linear models in multi-factor impact problems such as predicting housing prices. This is due to the comprehensive examination of multiple factors by nonlinear models, which linear models do not possess. The project still has many shortcomings in many aspects, such as not tuning the parameters of the SVM. There is also no addition of cross-validation, and a lack of visualization of the results.