Or: Data Mining for Everybody
Stumbling through the internet today and trying some new things, I began to realize the power of Google Trends and what it’s capable of showing. The unique situation of Google as a search monopoly has given it the position to monitor what the majority of people are looking for on the internet at any given time. The Hot Trends page shows recent events with an almost stunning similarity to news feeds and blogs, although searches aren’t intended to reflect what’s the news. This might seem like old news, with things like NPR proclaiming the power of Google in detecting flu pandemics and the like, but it didn’t fully hit me from things like that that these tools aren’t just available to researchers, the Google Trends page allows anyone to compare the prevalence of a search. Setting out on my own scientific endeavor, I looked for what G. Trends would have to say about Cake vs. Pie.
(cake = red, pie = blue)
Naturally I was vindicated to see that cake had search supremacy over pie, but a closer look yielded some interesting questions. What were all those little spikes on the pie graph? A look at data point A revealed it to be, of course, related to pumpkin pie for Thanksgiving! A separate trends search validated this, showing an extreme seasonal spike in pumpkin pie searches.
(Pumpkin pie)
Seemingly irrational spikes like that brought up more questions about the cake vs pie trends, and I noticed a weird spike that seemed to come up on the cake and pie graphs around what I assume is the spring line (1/4 way through the year). What could cause this jump in searches for delicious baked goods? Is there some sort of seasonal longing for pie and cake that crops up as spring comes? Unfortunately, there weren’t any data points that I could use to research this further, so the explanation for that little bump still lies shrouded in mystery.
The news reference volume is also interesting, particularly in so when concerning cake vs pie. Why are there more searches for cake, when there are more news stories tracked about pie? It could be some random use of the word pie in language of which I was previously unaware, but the difference between the two graphs still seems strange. My theory is that the insane media has become disconnected from the people, and obsessively carries stories about pie, when it is truly cake about which the populace wants to hear.
Language related searches, in addition, provide even more questions. Why do significantly more people search about pie than cake in Spain, defying the trend? Goodness knows.
Random spikes, dips, and otherwise weirdly outstanding patterns in search graphs is what makes Google Trends so powerful and interesting. Trying to correlate the irregularities with causes certainly proves to be a somewhat fun waste of time, even if it is mostly fruitless. There’s an endless stream of graphs to be analyzed (another interesting one is Yahoo vs Google – why are there more searches for Yahoo on Google, than Google on itself?), and it seems like each has some unique and funny graph to look at. Lastly, G. Trends always seems like its information hides some sort of deep hidden secret or potential – more than just tracking disease queries to a location, but imagine putting some sort of pattern detection to search queries and seeing what it might predict? The possibilities are endless, not to mention fun in a weird way, so just go play around with Google Trends for a while. It’s good times.
