I spent summer 2013 in Santiago, Chile, working for Urbana E&D, an urban design and investment firm. My work focused on the evaluation and expansion of their MEdIT, an econometric model of land prices in Santiago. The model utilizes data from real estate listings on portalinmobiliario.com and applies an urban economics framework which emphasizes the importance of travel time to employment opportunities and other amenities.
I incorporated data on the locations of roughly a dozen types of services and amenities, and used data mining techniques such as the LASSO and cross validation to identify variables important in predicting land value. I also tested the model for spatial autocorrelation and implemented a spatial error regression model to correct for it.