Real Estate Trend Prediction Using Linear Regression And Artificial Neural Network Techniques

An accurate assessment of future housing prices is crucial to critical decisions in resource allocation, policy formation, and investment strategies. In this work, linear regression and artificial neural network were employed to model home price indices, using datasets of the S&P/Case-Shiller home price index and twelve demographic and macroeconomic features in five metropolitan statistical areas: Boston, Dallas, New York, Chicago, and San Francisco. The data, ranging from March 2005 to December 2018, were collected from the Federal Reserve Bank, the Federal Bureau of Investigation, Macrotrends, and Freddie Mac. Three time-lagging situations were compared: no lag, a 6-month lag, and a 12-month lag. Since some data were available monthly, some quarterly, and some annually, two methods to compensate missing values, backfill and interpolation, were compared. The models were evaluated for accuracy and mean absolute error. The results showed that linear regression performed well in predicting long-term trends, while artificial neural network was suitable for short-term prediction. It was found that input factors that were statistically significant varied in different areas. The results also showed that the technique to compensate missing values and the implementation of time-lag influenced the models’ performances, both of which require further investigation.