a novel way to analyze this dataset! i don't think the color is doing anything other than showing that _any_ orders have ever occured in respective categories on the days of week this configuration would give you color by number of orders and aggregate those correctly for groups Plot.group( { fill: "sum" }, { x: "day_of_week", y: "name", fill: "orders", tip: true } )

i have a challenge: let's process the data so that names and sizes are separate, then keep the cell visual and show number of orders over a whole year (x=date). you can reserve FY for sizes or varieties, try both, see what works best. remove white gaps between cells and you'll get some kind of bitmap image where hopefully we'll see patterns

inspired by this https://observablehq.observablehq.cloud/framework-example-api/