Project: predict house prices

Project: price a house

Time to work like a machine-learning engineer: take a dataset and run a disciplined process from first baseline to a final, honest score. 300 houses, four features (size, bedrooms, age, distance to the centre), price in thousands.

1. Split the data three ways, define the score, and build a baseline.
2. Fit linear models, and engineer a better feature.
3. Compare with kNN, choose using validation data, then report the test score — once.