Housing Value Predicted Modelling using Random Forest Regression: Case study California Housing Dataset

Abstract

Housing price comes from many factors which are location, population, style of house, age of house, and people income. Many real estate developer companies use this data to predict price of house and give amount of investment for potential housing prices. In this study, we try to help the developer companies to predict price of house based on dataset. We try to build machine learning that can predict for housing price. There are three machine learning models that are used for this study, namely Linier Regression Modelling, Decison Three Regression Modelling, and Random Forest Regression Modelling. Each of those machine learning is trained using California Housing Dataset (1990) which is split into training set and testing set that training set contains 16512 instances and testing set contains 4128 instances. Training dataset is trained into each of machine learning model (Linier Regression, Decison Tree Regression, and Random Forrest Regression) after finished the training followed by evaluting the error prediction using K-Folds Cross Validation and showed by using Root Mean Square Error (RMSE). In this study, Random Forest Regression gives a better performance than two others (Linier Regression and Decision Tree Regression models) with error RMSE =49642.12.