Detrending

If we had the raw data..... That plea will recur throughout this exercise. ...we could see how many ways there are to detrend these data. Simplistic detrending of time series looks for a periodic component, with a 12 month cycle, or a 7 day cycle. Months are not as robust, due to their varying day numbers. But these data suggest that one could be much more refined. Many searches reveal a flurry of activity in Spring and Autumn/Winter, with a big dip in the Summer. This is, among other things, the academic year, and students do a lot of searches. Compare 'Jung' and 'Freud', for example:




You can see that Freud is taught in Universities, while Jung has gone out of fashion.

There might be other odd forms of recurrence. 'Breakfast' has a peak at weekends, but also in July.