ESCaPe 2020 Proceedings | Page 18

House Price Prediction using Random Forest Sankeerthan Kasilingam 1* , Anojan Satheesnathan 1 and Sampath Deegalla 1 1 Department of Computer Engineering, Faculty of Engineering, University of Peradeniya, Sri Lanka *E-mail: [email protected] Abstract: Housing price is influenced by multiple factors such as location, house size, number of bathrooms, and number of bedrooms. The traditional models used the statistics of these factors to predict house prices. The development of machine learning models to predict price as an alternative to the traditional model has been done in many countries. This had motivated us to do a study of houses located in Colombo district, Sri Lanka. The data set collected from the online house-selling platform is used for the study. The study focuses on removing outliers using K-means clustering and predicting price with Random Forest models. According to our results, after removing the outliers by clustering, the Random Forest model performance was improved and the RMSE of the best model was LKR 5.4 Million. Keywords - house value prediction, machine learning, regression, random forest 18