Cloud Hosted Ensemble Learning-Based Rental Apartment Price Prediction Model Using Stacking Technique

Abstract

In this chapter, online rental listings of the city of Hyderabad are used as a data source for mapping house rent. Data points were scraped from one of the popular Indian rental websites www.nobroker.in. With the collected information, models of rental market dynamics were developed and evaluated using regression and boosting algorithms such as AdaBoost, CatBoost, LightGBM, XGBoost, KRR, ENet, and Lasso regression. An ensemble machine learning algorithm of the best combination of the aforementioned algorithms was also implemented using the stacking technique. The results of these algorithms were compared using several performance metrics such as coefficient of determination (R2 score), mean squared error (MSE), root mean squared error (RMSE), mean absolute error (MAE), and accuracy in order to determine the most effective model. According to further examination of results, it is clear that the ensemble machine learning algorithm does outperform the others in terms of better accuracy and reduced errors.