MIT EECS Undergraduate Research and Innovation Scholar
Predicting Economic Shocks based on Consumer Credit-Risk Models using Machine Learning Algorithms
Andrew W. Lo
Customer spending and borrowing have always been a large part of the American economy, the former making up more than twothirds of the countrys GDP and the latter being a major contributing factor to the financial crisis of 2008. Consumer credit is a major market populated by credit cards, mortgage loans, and commercial banks. Patterns in customer borrowing can be analyzed and used to forecast potential shocks to the economy such as the 2008 crisis. However, due to the multifaceted aspect of decision-making, the large number of decisions involved in the consumer lending business, and the sheer amount of data available, building a reliable model over this information can only be done using efficient algorithms in machine learning and artificial intelligence.
I worked on a UROP with Professor Una-May OReilly on using genetic algorithms to produce visually appealing fractal-like images. I have also worked with Professor Szolovitz on optimizing Bayesian
network learning on datasets of genes. I worked at Apple Computers on building applications for Mac OS.