Melbourne Housing Price Predictor
Interactive Streamlit app estimating sold house prices for Ringwood, Bayswater and Nunawading using a Ridge regression model.
Overview
An educational Streamlit app that predicts sold property prices across three Melbourne south-east suburbs (Ringwood, Bayswater and Nunawading) from property attributes, backed by a Ridge regression model trained on historical sold-property data.
The problem
Wanted a simple, interactive way to explore how bedroom count, land size, property type and location interact to influence sold prices in a specific local market, without digging through raw sales spreadsheets.
My approach
Built a feature-engineering pipeline over a sample of Ringwood, Bayswater and Nunawading sold-property records - deriving total rooms, sale quarter, a suburb-property-type interaction term and a land-size-missing flag - then trained and compared regression models in scikit-learn, saved the best-performing Ridge Regression model with joblib, and shipped it behind an interactive Streamlit UI where users adjust property attributes to get a live price estimate.