Every time Daylight Savings Time ends in the US, every census record in the world (regional populations, nation statistics, etc) hiccups. For an example of what I am talking about, look at my population trend and zoom in on the first week in November 2016 or 2017. Every data point year-round is timestamped 03:00 until the Sunday when DST ends, where the timestamp is 02:00. The next data point has the same value, and the timestamp is 02:00 Monday. The next data point is normal, showing the increased population and a timestamp of 03:00 Monday. This hiccup creates an easily-visible flatline followed by a sharp jump. My guess is that the server runs the region update on schedule at 0330 EDT*, then the server time goes back one hour and the update runs again at 0330 EST, generating an unnecessary data point. The next day, the server corrects itself.
With the end of DST coming up in a couple weeks, would it be possible to fix this, and maybe remove the glitched data points from the existing records? It screws up my automatic data collection.
*Not the nation update, which runs from 0000 to 0100 Eastern. The "region update" is my term for the point where region averages and rankings are recalculated. I have observed this taking place at various times between 0330 and 0335 Eastern.