The date range is 9/22 to 10/8. That's two weeks, which means it's only 2 data points for each given day-of-week + time combo, so analysis might not be accurate for DOY charts.

Otherwise, since the sample size is large (2000+ pts in the worst case), this is a good analysis.

How are you making the charts? It's not any JS library I've seen, and looking at the source, the code appears to be custom.