As a facet of Alveo, the Autotrader machine learning tools were underused by traders and undersold by account managers despite their potential to offer VMM a unique advantage in the marketplace. The output of the Python-based machine learning algorithm was inscrutable to everyone in the company without a background in mathematics.
My first step was to present the basic output in a simple, structured manner that matched the existing UI.
I then worked closely with our data analyst to brainstorm ways to make the results meaningful and exciting for the rest of the company and our clients. We began playing with area as a way to represent value, as well as ways to show the relationship between the levels of filtering. I rapidly produced visual mocks in Illustrator to consider these relationships.
We ultimately concluded that the key value of the tool was in the ability to drill down with more specificity than a trader could do manually in the same amount of time.
I focused our prototypes on models that would show increasing specificity over time. I continued to work closely with our analyst to curate meaningful data points and a strong narrative flow.
I then went straight to a code prototype using a sample client dataset selected by our data team. The interactive visualization was ultimately used for both training members of the company and for pitching the tool to clients.