House Price Prediction in HCMC
Scraped 100,000+ listings from Nha Tot, trained a model that tells whether a house in Ho Chi Minh City is in the high or low price range, and put it online as a web app on AWS EC2.
Problem
Predict the price range of a house in Ho Chi Minh City from its features, using real listings instead of a sample dataset.
What I did
- Scraped more than 100,000 listings from Nha Tot with Selenium and Python.
- Cleaned the data in Pandas and Google Sheets, and engineered features to improve the model.
- Trained a Scikit-learn classifier on location, size, amenities and other features.
- Deployed it as a web app on AWS EC2 so people could try predictions in real time.
Result
79% accuracy overall, with an F1 of 0.84 for the high-price class and 0.82 for the low-price class. I was the team leader.