Home Sale Price Prediction in Boston, MA: UPenn MUSA507
(This fictional report was prepared as a class assignment for MUSA507 at the University of Pennsylvania)
My client, Zillow, has realized that the predictions of home prices in the Boston area which are currently on the site are not sufficiently accurate. This lack of accuracy is putting the site's credibility in question and hurting its ability to attract and retain customers in the region.
Accurately predicting home prices is difficult. Home prices are a function of many complex considerations including the physical characteristics, location, and other factors including seasonal variations and the personal preferences of buyers and sellers.
To achieve better prediction accuracy, I have collected data to capture as many of the factors listed above as possible. More specifically, I worked to capture the variation in prices due to both physical characteristics of homes, and the underlying spatial patterns of prices in the region.
The resulting model was able to improve predictions considerably, with a new error margin of only about 11%-12%.
Please click on the link by the repo title to see full markdown.