S39 A Python-Based Approach to Improving the Purdue Forecast Game

Sunday, 6 January 2013
Exhibit Hall 3 (Austin Convention Center)
Daniel G. Burgin, Purdue University, West Lafayette, IN; and M. E. Baldwin, D. Carroll, L. C. Dawson, K. Hoogewind, D. Snyder, A. J. Stepanek, J. M. Woznicki, and Q. Zhu

Since the 1970's, Purdue University students have managed a forecasting contest called "The Purdue Forecast Game," allowing fellow students, faculty and alumni to showcase their forecasting skills in friendly competition. However, the forecast evaluation process has become quite burdensome with substantial manual labor involved and periodical delays in forecast and verification updates, ultimately resulting in decreased interest amongst participants. As part of a group project for a course in forecast verification, our goal is to automate specific tasks including data collection, scoring and visualization of forecast performance with the added benefit of increasing involvement and enhancing learning. The Python software package was utilized as a tool to streamline the process. The result of this work is a Python package that will be used to maintain and improve the Purdue Forecast Game.
- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner