Estimating the market value of a home is a task on which buyers, sellers, lenders and agents still rely heavily on informal judgment. This research project develops and evaluates a web-based House Price Prediction System in which supervised regression models are served through a Django application. Users enter six property attributes (built-up area, bedrooms, bathrooms, year of construction, urban or rural location and garage availability) and receive an instant estimate, while administrators manage users, review stored predictions and produce date-range reports. Linear regression, decision tree regression and random forest regression were trained on a controlled dataset of 2,000 records and compared on a held-out 20 percent test split and with five-fold cross-validation. The random forest gave the lowest errors, with a test coefficient of determination (R-squared) of 0.997 and a mean absolute error of about 0.23 million rupees, whereas the linear model reached 0.930. Feature-importance and ablation analyses show that predictions depend mainly on area and location, and a scalability test shows that the serialized forest grows linearly with training size. Because the data are synthetic, the reported accuracy shows that the pipeline works as designed and should not be read as real-market performance; limitations and the steps needed to move to real transaction data are discussed.

M Angel T Maria Ruban, Dr. Manoj Singh, "House Price Prediction System Using Machine Learning" International Journal of Scientific Research and Engineering Development, V9(5): Page(1474-1481) September - October 2026. ISSN: 2581-7175. www.ijsred.com. Published by Scientific and Academic Research Publishing.