Need best practices for visualizing AI data limitations.

Screenshot 2018-02-13 15.54.29

Part of understanding and trusting a machine learning result is understanding the limitations of the data that was used to create the model.

Nathan Yau provides an excellent description of Visualizing Incomplete and Missing Data for descriptive analytics.  His post presents the best practices in the field.

To achieve our XAI goals we need be as thoughtful about communicating to users the limitations of the data that is driving predictive and prescriptive analytics that flow from our models.

So here is a challenge for the community – who is going to define best practices for visually communicating the limitations of machine learning training data?

 

Leave a Reply