Power BI is great at aggregating data, but when you have many data points and want to visualize their distribution, this can be challenging.

Kernel density estimation (KDE) is a means of uncovering the distribution of your data (think: "smooth histogram"), and this is commonly seen in violin plots to visualize the results. There are custom visuals that offer this functionality, but these have a limit on the number of rows that you can provide. If you have an extensive data set (hundreds of thousands or even millions of rows), you can only obtain so much data and can't get a true representation of your distribution.

In this session, we will review the challenges of analyzing a large dataset with a custom visual and talk through how we can write our own KDE function using DAX, which will quickly calculate the distribution of a vast data set, and then also use DAX to turn the output into SVG for display in a native visual.