House Price Prediction Using Machine Learning Techniques

Abstract

Prediction of housing market prices is critically important because many sectors of the housing market rely on accurate house price forecasts to make decisions about investments and financing. Therefore, this study presents an extensive analysis of the prediction of residential property prices in Delhi through using a machine-learning approach to connect multiple sources of information from the housing sector, including a curated database of housing sales in Delhi, geographical locations such as locations to metro stations, schools, hospitals, etc., and local environmental quality (i.e., air quality index). Through feature engineering and through the use of a Random Forest regressor trained with 7738 records of housing sales... The Random Forest regressor trained on 7738 records of housing sales has provided good performance with an R² of approximately 0.92 when tested against the test dataset when forecasting sale price for individual residential properties. Res1lts show that including data about location-related amenities (such as subway stations) and air quality significantly increases predictive accuracy of the sale prices of the houses in Delhi. An description of the models' output and analysis is also provided, as well as the implications of the findings for urban planning and housing market analysis.