What-if Tool
What-if Tool¶
A key challenge in developing and deploying responsible Machine Learning (ML) systems is understanding their performance across a wide range of inputs.
Using WIT, you can test performance in hypothetical situations, analyze the importance of different data features, and visualize model behavior across multiple models and subsets of input data, and for different ML fairness metrics.
This video shows the setup of the What-If Tool, including the notebook environments it works in and how to use it with Cloud AI Platform models. Learn how to integrate the What-If tool with Google Cloud and help ensure your models treat all users fairly:
In the following video, Google’s Yufeng Guo speaks to how you can inspect your machine learning models via the What-If tool. Specifically, he speaks to the different features that are integrated into the What-If tool - such as the Facets Dive tool, Performance & Fairness tab, and the Facets Overview tool - that allows users to create predictions for new edit values, check overall model performance, and look at the distribution of data points within their models:
Resources