scam_watcheroo wrote:
If we simply plot Mimi's raw rpm cadence data in a histogram at 1 rpm intervals, we can see the is a clear pattern that has so far defied all my attempts at reaching an explanation for it.
...
Can anyone explain this to me?
Seems to me like a very clear repeating pattern of gap, 1x count, 2x count.
This is the type of bias that would result from always rounding half-numbers one way vs. the other. The simplest explanation of this would be that data is rounded to integer RPM from some calculation that produces fractions, and is consistently rounded in a way such that every third number rounds down (vs. up). So 59.5 (round up), 60.5 (round up), 61.5 (round down), 62.5 (round up) could produce the pattern you see.
Furthermore, this type of error would only occur on the finest grain setting and would disappear with smoothing internally or externally to the watch.
I have zero knowledge of how cadence is calculated, simply tossing out the obvious if you want to try to dig into Suunto internals to further analyze.