Here are some images for different periods of the sine wave per day. Note that I have maintained the samples at 15min time intervals.
The first is for 1.5 full cycles per 24 hours.
The second is for 0.75 full cycles per 24 hours.
It’s tempting to still see patterns in the confidence for these images. In particular the predictions seem to have better confidence just after the peaks/troughs and worse near 0. This is consistent with there being a higher density of points near the peak/troughs than near 0.
They’re fresh off the press, so I haven’t decided how to interpret them properly yet.