Note that I still include the empty blocks for the red-line indicator so it's easy to see with and without that calculation.

The block fullness is actually fairly variable so the averages you get depend very much on your sampling time. Plus the number of blocks in an hour is also somewhat variable. You can have 4 blocks one hour for 98% full and then maybe 10 blocks another hour for an average of 40% full. The problem is, those 10 blocks at 40% don't really help the person who couldn't get their transaction on the chain in the 4-blocks-per-hour time period.

As always, the question has to be, "what am I looking to represent". A daily average might not present a useful number if you're wanting to know about user experience whereas an hourly can give a better insight as that's closer to the timescale for a typical Bitcoin transaction. It doesn't help that the daily average interstate speed was 50mph when an accident means you're stuck in non-moving traffic for an hour.

I haven't looked but transactions may be on a 24 hour cycle also. Low transaction volumes at certain times of the day would contribute to a low daily averages while high volume times may still be having problems that would indicate a need for remediation.

In summary, averages are tricky when it comes to interpretation and divining meaning. Which is why I included the per-block indicator dots.