Explainable and Unbiased Predictive Analytics for Institutional Researchers

Session Abstract: With the rapid growth of analytical methodologies from both statistics and data science, the ethical and transparent application of algorithmic outputs in support of university operations and activities has become increasingly important and has generated concern in the higher education community. This session discusses effective participatory approaches that ensure actionability of analytics, support the production of explainable Artificial Intelligence and Machine Learning, and minimize and effectively address the risk of bias in advanced analytics. The session will close with an interactive discussion on what it means to provide effective, accessible, and unbiased application of advanced analytics in institutional research.

Keywords: explainable AI and ML, bias in advance modeling, actionable analytics, participatory action research

Contributing Authors: Jesse Talley – IUI Campus Manager of Institutional Effectiveness, Indiana University Indianapolis; Christina Downey, IUI Campus Manager of Institutional Effectiveness – Associate Vice Chancellor for Undergraduate Education and Dean, Indiana University Indianapolis