The data I have is in the form of customer receipts. Items are grouped into categories such as Dairy, Produce, Meats, etc ...
I am putting together a dashboard of information, one plot of which is the daily total amount of purchases of each category as a $ amount. I have a 7-day rolling average overlayed on this which makes it much easier to interpret.
Is it valid and useful to convert this type of plot into a percentage and overlay a rolling average to see the trend? For example, I would like to see how much of total purchases per day are Dairy. If on Day 1, there was \$50 spend on Dairy and \$150 otherwise, it would be 25%. If Day 2, no one bought Dairy, it would be 0%. If Day 3 came along and only $500 of Dairy, it would be 100%. Since this plot is potentially very spiky, I wanted to smooth it out for easier interpretation of the trend.
I've seen posts against unweighted averaging of percentages, but I would like to check if this is a valid use case.