Detecting and Interpreting Variable Interactions in Observational Ornithology Data

In this paper we demonstrate a practical approach to interaction detection on real data describing the abundance of different species of birds in the prairies east of the southern Rocky Mountains. This data is very noisy - predictive models built from this data perform only slightly better than baseline. Previous approaches for interaction detection, including recently proposed algorithm based on Additive Groves, might not work ideally on such noisy data for a number of reasons. We describe the issues that appear when working with such data sets and suggest solutions to them. We further demonstrate that with our improvements to the interaction detection algorithm it is possible to detect interactions between important features and the response function, even when the data is this noisy. In the end, we show and interpret the results of our analysis for several bird species.

• Type: Conference
• Year: 2009
• Where: ICDM
• Authors: Daria Sorokina, Rich Caruana, Mirek Riedewald, Wesley M. Hochachka, Steve Kelling
• Keywords: Data Mining, ICDM 2009, Interaction Detection, Noisy Data, Southern Rocky Mountains