Are you looking to buy a house? The chances are, if you don’t already own a house, you are or you will be some time down the road. For this project you will build a regression model for house prices. How cool is that! You will have a bargaining power against the property agent. Instead of following blindly the price quote, you will have a better reference for whether the house you want to buy is overpriced or underpriced.
Brain Storm: what factors do you think would affect house price?
May be the location (the district – a variable coded by zipcode)? The square feet of the house? The age of the house? Whether it has a swimming pool? Whether it is foreclosed? Feel free to do research on the Internet and come up with a list of possible variable candidates.
Data collection: use Zillow.com
Make sure you collect data on more than 30 houses. Make sure your sampling (if sampling is used) is random. Think about it: how can you ensure randomness?
Does any of your predictor turn out to be significant? Use the model to predict the sales price for the house you are most interested in. Is that house overpriced or underpriced? Are you getting a bargain?
In your report, be clear about:
- How do you come up with the predictors?
- How do you collect data? If random sampling is used, how do you conduct sampling? What mechanism you used to ensure randomness?