This graph plots the observed rate of tweets for Eric Fischer’s Twitter feed over the period from January 8, 2015 through November 30, 2015.

This graph uses a time scale to plot time-of-day. Time scales are normally used to plot absolute time: a specific moment on a specific day in a specific year. Here, though, we’re interested in studying the daily pattern, so data from many days is collapsed onto a single day. (That day happens to be January 1, 1900, but the choice is arbitrary as long as we’re consistent.)

One wrinkle here is that the data is stored in UTC, but we want to view the pattern in Eric’s local time. To do this, we offset the time by seven hours, since the majority of the data was collected during Pacific Daylight Time (-0700). However, this is incorrect because some of the data was collected during Pacific Standard Time (-0800). Since the local time zone offset changed during data collection, the pattern is somewhat blurred. To fix this, you would need to aggregate the data by local time-of-day instead of UTC time-of-day.