# The What-If Tool: Interactive Probing of Machine Learning Models

## TL;DR

The paper introduces the What-If Tool, an open-source interactive application enabling practitioners to probe, visualize, and analyze machine learning models' performance, fairness, and feature importance with minimal coding.

## Contribution

It presents a novel, user-friendly tool for ML model analysis, supporting hypothetical testing, feature importance, and fairness evaluation, enhancing interpretability and deployment insights.

## Key findings

- Facilitates understanding of model behavior across diverse inputs
- Supports analysis of multiple models and data subsets
- Enables measurement of fairness metrics in ML systems

## Abstract

A key challenge in developing and deploying Machine Learning (ML) systems is understanding their performance across a wide range of inputs. To address this challenge, we created the What-If Tool, an open-source application that allows practitioners to probe, visualize, and analyze ML systems, with minimal coding. The What-If Tool lets practitioners test performance in hypothetical situations, analyze the importance of different data features, and visualize model behavior across multiple models and subsets of input data. It also lets practitioners measure systems according to multiple ML fairness metrics. We describe the design of the tool, and report on real-life usage at different organizations.