Using Supervised Machine Learning to Predict House Prices

Abstract

Given the recent influx of prices in the housing market, determining a fair housing price has been of high interest for many homebuyers and sellers alike. In this project, various machine learning models are used to predict the price of a house based on physical features and characteristics such as lot size and neighborhood. Extensive data preprocessing and feature engineering were employed to aid the models’ performance compared to other models in the market. The best models have been able to predict U.S houses’ prices within a RMSE value of $23,000 when the mean price of a house in the dataset is $180,000. In future research, this model can be implemented in various other places within the U.S and additional features can improve performance further.