While you were celebrating new year's, the rest of the country was busy ordering Domino's pizza! Facts for your perusal:
Hottest searches in India on 30-Dec-2010 (via Google) – No Domino's
Hottest searches in India on 31-Dec-2010 (via Google) – "Domino's India" on #13
Hottest searches in India on 01-Jan-2011 (via Google) – "Domino's India" on #13
Hottest searches in India on 02-Jan-2011 (via Google) – No Domino's
(That most of the country was also searching for SMS spam to send on New Year is a separate story!)
Now take a look at the weeks for which "Domino's" related searches have peaked – correlates pretty well with major Indian festivals.
A cursory glance at other fast food chains, like McDonald's, Pizza Hut, & KFC also reveals a peak around the new year. However, nothing seems to rock the Indian palette on a festival as Domino's pizza! I, too, had ordered Domino's pizza on Christmas eve last year. What about you? Hungry kya?
Update: And the trend continues. Check out the huge spike in search volume for "dominos india" on Google Insights in the week of Holi (20-26 March, 2011)
Since your last two posts have been about drawing inference from single points of statistical data, I had to warn you that it's very possible that your inference might be wrong. (Need not be, but you can never be sure.)
ReplyDeleteThis is due to an interesting phenomenon called "Simpson's Paradox" which basically says that as you get more variables in the data-points the inferences you've drawn before might not correlate with the inferences you've drawn since.
Here's a better explanation: http://lesswrong.com/lw/3q3/simpsons_paradox/
I guess unsureness isn't really a problem with blog posts. They do make for interesting discussions (however useful or useless) :-)
ReplyDelete@Rakesh: That's an extremely interesting phenomenon. Never really thought about statistics & inferences in that way. But, sometimes doesn't thinking too much lead to "analysis paralysis"?
ReplyDelete@Anonymous: Glad to be seeding some discussions.
ReplyDeleteHere is more interesting possibility due to Simpson's Paradox - it is in theory possible for Sachin and Sehwag to play 10 tournaments, where VS has a batting average that is higher than SRT in each of the tounaments and yet, SRT can have a higher overall average than VS in these 10 tournaments.
ReplyDeleteIt's a little more intuitive here since the overall average depends on the number of matches played for each of the tournaments and things can be skewed if SRT plays only one match in a few low scoring tournaments and all the matches in high scoring ones (with Sehwag doing the reverse).