The Comparison of Six Prediction Models in Machine Learning: Based on the House prices Prediction

Abstract

There are many different kinds of prediction models that have different performances when faced with different kinds of data. This essay focus on the comparison of the performance of multiple regressions, SVM with RBF kernel function, and Random Forest when predicting the house prices of Boston and using score function, k-fold cross-validation and shuffle cross-validation to evaluate their performance respectively. Finally, parameter adjustment, grid search, and forward selection are introduced to improve their performance. By combining the result given by three evaluating methods, SVM with RBF kernel function is the better model and the Random Forest is the worst one, whose scores are higher than 0.7 and lower than 0.1 respectively. And all of these three functions can slightly improve the performance, especially, the effect of the grid search is the best one, which can improve the score by 0.023 higher than the original score.