Using Linear Regression and Machine Learning Techniques to Predict Housing Prices Based on Economic Factors

Abstract

When trying to predict housing prices, most studies rely on data from a specific area, and the features of the homes there. In this study, the goal is to use linear regression and machine learning to predict housing prices based on overarching economic factors. A mix of machine learning and linear regression was used, including TensorFlow Keras, OLS, Ridge, Lasso, Elastic Net, XGBoost, Random Forest and SVM. Datasets featured include Average Sales Price of House Sold for the United States, closing stock prices (NASDAQ, S&P), 30-year mortgage interest rates, average monthly rent, number of houses sold, number of houses constructed, mean family income, median family income, GDP, and unemployment rate.