Your math may be fine. But your method is invalid, you can't learn what you need from that.
I'd give it a 2% chance that you need more than 200A.
The POWER working group and Lawrence Berkeley National Lab have instrumented tens of thousands of homes, and analyzed large volumes of utility data. The number of homes that require 400A service is tiny.
The chart is odd, but well founded. It shows the distribution of homes by peak load.This is a very weird chart. I'm not sure why I need to see the total quantity of panels on the X axis. It would however be interesting to see quantity of each panel size, or quantity of each utilization percentage.
Regardless, it is making me question some 400 amp services on recent and future project.
I think the X-axis is the Identifier for a single home, not the number of homes at each data point.
It is needed to create the data point. It appears they sorted the houses by service size and usage, then assigned an identifier so they could be plotted individually.
I had assumed it was simply the quantity based on the total panel count of 1,480. It could be one in the same as what you're describing. Either way it seems useful to the researchers, less so for others hoping to glean useful information from the chart.
As various commenters noted: its a way of visualizing the distribution of homes.
The quantity of 100 amp panels is 407 (482 minus 75). Over half of them (200 of roughly 400) never saw a peak over 50% utilization on a 15 minute basis, during the sample period.
I can't imagine I would have anything of use to add, way over my pay grade. I was merely commenting on what I thought was a poor choice on the layout of the chart. I'm guessing that chart is but one of many in an otherwise well researched and informative study.
Both seem relevant, useful and interesting.