Regression Models with Different Levels of Complexity for Real Estate Price Prediction

Real estate decisions are typically among the things that most people give careful thoughts to before making. However, the most difficult part of picking which real estate to invest in is narrowing down on which property is best considering the abundance of factors. That is why this study is aimed to see if different regression models such as linear regression, random forest regression, and extreme gradient boost regression are viable in helping people pick out the right properties. Finding and preparing a dataset of housing features and pricing would be the first step in the experiment. The process proceeds with training the models in specific preset environments. The experiment resulted in the random forest regression performing the best with an R
²
score of 0.7375 in the 80:20 data split and the linear regression performing the worst with an R
²
score of 0.4149 in the 70:30 data split. To add, the fact that all 3 models behaved in a consistent manner across all environments is noteworthy. Despite the findings, we believe that the research could benefit from trying other models or changing the dataset for better results.