Using Visual Analytics to Interpret Predictive Machine Learning Models

TL;DR

This paper explores how visual analytics can interpret black-box machine learning models, enabling understanding of their reasoning without compromising predictive accuracy, through identifying solution spaces and practical examples.

Contribution

It introduces a framework for using visual analytics to interpret predictive models without reducing their accuracy, supported by real-world applications.

Findings

Visual analytics aids interpretation of black-box models.

Interpretability can be achieved without sacrificing predictive power.

Practical examples demonstrate effectiveness in real-world scenarios.

Abstract

It is commonly believed that increasing the interpretability of a machine learning model may decrease its predictive power. However, inspecting input-output relationships of those models using visual analytics, while treating them as black-box, can help to understand the reasoning behind outcomes without sacrificing predictive quality. We identify a space of possible solutions and provide two examples of where such techniques have been successfully used in practice.