#9: Macroeconomic Forecasting using Analogy Weighting in Stan

Jim Savage will present Analogy Weighting, an easy-to-deploy method that helps modelers train time-series models on the most relevant data. In doing so, the method frees modelers from the question of what the most optimal look-back window should be, while typically improving out-of-sample predictive performance. He will present applications to forecasting macroeconomic variables, as well as portfolio volatility during the 2008-9 crisis.

Jim is an applied economist, data scientist, and head of Modeling at Lendable, a frontier markets finance start-up. He recently moved to New York from Australia, where he worked at the Grattan Institute, a think-tank, and at the Australian Treasury. Last year, he was a fellow in the University of Chicago's Data Science for Social Good summer fellowship (2016 applications close Feb 1!).

(Please, register with your full name as we will need it for the ISE building security)

http://photos3.meetupstatic.com/photos/event/a/9/e/600_445862718.jpeg