Housing Price Prediction Using Machine Learning Techniques

This research employs advanced machine learning algorithms to predict housing prices, specifically Linear Regression, Random Forest, Gradient Boost Machine, Multilayer Perceptron (MLPs), and Long Short-Term Memory (LSTM). The study aims to address the oversight of crucial house factors such as front area, location, size, and others, to provide effective housing pricing predictions. The methodology involves structured processes, including data collection, preprocessing, feature extraction, dataset splitting, and the implementation of machine learning models. Using a housing prices dataset from Kaggle, the study explores attributes like sale price, lot area, street type, and more. The dataset, comprising four files (Data description, sample submission, test, and train), consists of 163 columns, each record encompassing 80 attributes that describe specific houses. The study emphasizes the significance of data preprocessing involves handling missing values, outlier removal, and log transformation for improved results. The models' performance is evaluated using metrics like Mean Squared Error (MSE), Root Mean Squared Error (RMSE), and R-squared (R
²
) on the test dataset. The results indicate that the models achieve promising outcomes in predicting housing prices, with Linear Regression leading with a remarkable accuracy of 91%, followed closely by Gradient Boost Machine (0.911) and Random Forest (0.905). While MLPs score lower at 0.584, LSTM provides a commendable R
²
score of 0.863. The research contributes valuable insights for accurate predictions in real estate, considering factors influencing house prices.