In 2021, the software platform that began as a vehicle for client solutions at White Whale has made huge strides. While there is a great deal more to come, we decided it was time to put the pieces together and share some of our analytics platform's latest UI and UX improvements with the world. Over time, we've found that there is no better way to describe what DeepSea can do than by creating an example in context. For inspiration, we turned to a sport beloved by many of the White Whale team; one with a globally-distributed, dedicated fan base; one which has, by nature, a wealth of data.
We built a public dashboard hosted on DeepSea, our web-based AI analytics and dashboarding platform, to visualize the results and descriptive statistics of the 2021 Formula 1 season. The wealth of F1 data makes it a great subject for a fun side project like this because of the multiple lenses through which to look at the data: season vs. race; driver vs. constructor; various different types of data and ways of grouping them, necessitating multiple different chart types to visualize the data.
We wanted to use DeepSea here the way we use it for clients: to identify and highlight patterns.
DeepSea has been developed within a unique, user-centric design approach. This makes it easy and intuitive to create, resize, and place visualization elements within the dashboard. These could be a KPI, table, or chart of any of the following types: area, line, scatterplot, bar, pie, radar, or boxplot. In this case, our data scientists built charts to visualize drivers' and teams' cumulative points by round of this season, a distribution of average pit time, total wins by team, lap times by driver over the course of each race, and so on, by dragging and dropping these variables into the chart-mapping interface for DeepSea to automatically build. The drag-and-drop chart building interface brings many functionalities to simple point-and-click commands, like choosing variables to plot, applying operations to them (sum, average, etc.), filtering for certain values, and grouping or ordering them automatically.
Another unique component of DeepSea's interface which makes complex actions simple for users with a wide range of technical expertise is our model onboarding UI. DeepSea is not just a platform for data visualizations– our data scientists, in this example, are able to upload custom code and customize the parameters through DeepSea's model onboarding UI. This script pulls F1 data updated after each race, and the dashboard is refreshed to reflect the most recent data.
DeepSea's dashboard infrastructure allows for adding, duplicating, and deleting multiple tabs within a workspace. In this example, we created a Season tab and a Race tab. Users can view the Season dashboard for an overview of the current F1 season, scroll through and easily toggle on or off individual data points for drivers, teams, or races to hone in on specific points of interest. They can also filter the Race dashboard for any particular race this season to paint a picture of how it went, filtering individual cards for certain data points they're looking for. For example, the user could possibly filter to exclude anomalies from the data for a more accurate representation, or index the final lap for accurate values of the drivers' finishing positions.
The mission behind DeepSea has always been to bring the insights from machine learning models to the necessary users and decision makers in a way that’s accessible. Great visualizations are the best tools for this, highlighting patterns and outliers and using colour intuitively to tell a story. The concept of empowering users of all levels of technical ability to work with and understand their data has been a central part of White Whale’s philosophy from the beginning – and we can’t wait to show you where else this has taken us.
