Property Price Prediction Using the Random Forest Regression Algorithm

Abstract

This study uses an innovative approach to predicting property prices using machine learning with the Random Forest Regression method. The dataset was obtained from Kaggle and consists of 500 rows with 12 attributes, comprising 10 numerical attributes and 2 categorical attributes. The evaluation results, calculated using the R² score on the test dataset, show strong performance, achieving the highest R² score of 81.88% with a dataset split ratio of 90:10. The scatter plot visualization indicates shows the model's predictions often correspond closely with the actual values, showing strong accuracy, despite a tiny gap between the anticipated and real values. The graph comparing the training data and the actual data shows no significant signs of overfitting or underfitting, demonstrating the Random Forest Regression model's strong accuracy in predicting house prices and its capacity to effectively capture the relationship between independent and dependent variables.