Interpreting Interpretability: Understanding Data Scientists’ Use of Interpretability Tools for Machine Learning

CHI 2020 |

CHI 2020 Honorable Mention Award

Machine learning (ML) models are now routinely deployed in domains ranging from criminal justice to healthcare. With this newfound ubiquity, ML has moved beyond academia and grown into an engineering discipline. To that end, interpretability tools have been designed to help data scientists and machine learning practitioners better understand how ML models work. However, there has been little evaluation of the extent to which these tools achieve this goal. We study data scientists’ use of two existing interpretability tools, the InterpretML (opens in new tab) implementation of GAMs and the SHAP Python package. We conduct a contextual inquiry (N=11) and a survey (N=197) of data scientists to observe how they use interpretability tools to uncover common issues that arise when building and evaluating ML models. Our results indicate that data scientists over-trust and misuse interpretability tools.

Furthermore, few of our participants were able to accurately describe the visualizations output by these tools. We highlight qualitative themes for data scientists’ mental models of interpretability tools. We conclude with implications for researchers and tool designers, and contextualize our findings in the social science literature

Publication Downloads

InterpretML

July 16, 2019

InterpretML is an open-source python package for training interpretable models and explaining blackbox systems. Historically, the most intelligible models were not very accurate, and the most accurate models were not intelligible. Microsoft Research has developed an algorithm called the Explainable Boosting Machine which has both high accuracy and intelligibility.