Graduate Seminar | Towards Enabling Understanding and Interpretability in Data Analysis

Abstract

Data analysis is used widely across domains, often for decision making tasks. However, one of the central problems in data analysis is the lack of interpretability and understandability of results. These methods rarely provide insight into the internal operations that led to the results and are often classified as “black box”.  My research focuses on employing visualization to illuminate behaviors in analysis methods and provide additional insights into the final results. In this talk, I will demonstrate my efforts along two avenues: 1) improving interpretability of dimensionality reduction plots, and 2) enabling understanding in general analysis methods. 

***********************************************