Strong inexplicable periodicity
52 views
Skip to first unread message
Nonsuch
Feb 17, 2010, 11:45:46 AM2/17/10
to Google Trends
I suspect there is a flaw in the algorithm Google Trends uses to
calculate its values, which causes it to produce a spurious seasonal
periodicity for certain kinds of data.
Try "searches" for the two terms sustainable, biodiversity - you'll
see a coupling of the two trends that is so highly similar over the
period 2004-2009 that climate change modellers would hand over eye
teeth.
If you're not convinced, then extract the data and calculate the mean
for each of week 1... to week 52 and plot them for both these series.
The seasonal trends are close to identical - not just in phase, but
also in amplitude.
Google Trends, what's going on?
calculate its values, which causes it to produce a spurious seasonal
periodicity for certain kinds of data.
Try "searches" for the two terms sustainable, biodiversity - you'll
see a coupling of the two trends that is so highly similar over the
period 2004-2009 that climate change modellers would hand over eye
teeth.
If you're not convinced, then extract the data and calculate the mean
for each of week 1... to week 52 and plot them for both these series.
The seasonal trends are close to identical - not just in phase, but
also in amplitude.
Google Trends, what's going on?
Reply all
Reply to author
Forward
0 new messages