Forecasting Multan estate prices using optimized regression techniques

Purchasing a house or plot has become a complicated task for an average person due to budget constraints and market situations. An individual does not know the prices of the plots and gets trapped by a middle man. This paper proposes a solution for this problem by predicting the plot prices using a machine learning approach, leveraging Multiple Linear Regression, Gradient Boosting Regression, and Random Forest regression techniques. This work compares the performance of these three algorithms by hyper-parameter tuning using gird search, and random search for checking which one is adequate in terms of scores and error rates. Factors that influence the prices of the plots include plot covered area, physical condition of the plot, area, and population. Gradient boosting regression has surpassed all other machine learning methods, achieving the lowest error rates and highest R-squared score of 0.987 with grid search. The resultant predictive systems can help the folk in three ways. 1) safety from deception 2) budget oriented instant information, and 3) time saving.