House Valuation API

Quick Info / Links

In this project, I analyzed a house listing dataset from a competition on Kaggle (sponsored by Zillow). Although the competition had ended long before I found it, the data was interesting and useful enough for my own independent project. I decided to clean/analyze the dataset, and use it to experiment with various machine learning models aimed at predicting house sale price. The Jupyter notebook linked above contains and thoroughly explains all of my work in these cleaning/analyzing/experimenting phases. I ultimately settled on a CatBoost gradient boosted tree model (leading to fairly accurate predictions, considering the complexity of the target feature). I deployed the model via a simple API (linked above). Try out the demo to see how much your house might be worth! (Unfortunately Zillow has obfuscated the location data, so you'll just have to choose arbitrarily from the county, city, and zip code options.)