Enhancing Real Estate Price Forecasting Using Advanced Machine Learning Algorithms: An Empirical Evaluation with Real Market Data

Abstract

Predicting real estate prices accurately has become crucial for buyers, investors, and legislators to make wise choices. A viable strategy for identifying the complex patterns and trends in an ever-changing market is to use cutting-edge machine learning algorithms. The purpose of this study is to offer reliable conclusions about the dynamics of the Lahore, Pakistan, real estate market, especially as it relates to inflation and other economic shifts. The goal is to improve forecasting techniques in order to promote a real estate market that is more open and effective.  The results are intended to assist in making decisions and to provide guidance for policies that encourage steady, sustained market growth.  To forecast housing prices, a number of parametric regression models were used, such as the Extra Trees Regressor, XGBoost, Random Forest, Gradient Boosting, Decision Tree, and CatBoost Regressor.  As of June 26, 2023, data set was collected from a public website. Data set comprises on 9,539 listings from 6 districts. Additionally, As of June 26, 2023 in Pakistan real estate market, the average inflation rate in 2024 was 24.76%. The models that performed the best among the ones that were assessed were Gradient Boosting and Extra Trees Regressor, which had the lowest mean squared errors (MSE) and R² scores of 85%. Additionally, CatBoost shown competitive performance and is emphasized for its usefulness. The study emphasizes the importance of particular property attributes in predicting prices and advances our knowledge of machine learning applications in real estate forecasting.