House price prediction in Dhaka using Machine Learning Algorithm

Abstract

This study investigates the application of various machine learning models for predicting house prices using a dataset sourced from Bproperty, which includes 598 entries and seven attributes: Address, no. of bedrooms, no. of bathrooms, area sqft, type, features, and price. The methodology encompasses several key steps: It includes data selection, data cleaning where missing value removal is done, feature selection, encoding, exploratory data analysis EDA, model training, model evaluation and model testing. The analysis compares the performance of four regression models: Polynomial Regressor is one among them, SVR (Support Vector Regression), XGB Regression, and Random Forest Regression are the other models. Based on these models the R² score was calculated to be 27. 03% for Polynomial Regression: 40. 14% for SVR; 97. 16% for XGB Regression, and 97. 73% for Random Forest Regression. These outcomes suggest that the techniques like XGB and the random forest outcompete other regression models for efficiency of prediction. This probably explains why Random Forest and XGB are able to make much better predictions of the data since they have the capacity to model interactions that are non-linear in nature. These results reveal an important need for the usage of modern techniques in machine learning for reasonable and accurate house price estimate important for the stakeholders involved in the real estate market.