The correlation between rain and depression is interesting!

I really like the diet trend and how it dramatically drops in the last quarter of the year and sharp increase at New Years! I think it is less to do with Thanksgiving and more to do with New Years resolutions, but still extremely interesting!

Notice the little bump in October before the spike in November? That's right kids, Canada celebrates thanksgiving the first week in October.

Very amusing. I wonder if weight watchers gives out free turkeys to boost sales.

As the slashdot crowd like to say, "correlation is not causation".

It's likely that "rain" correlates with the seasons, and "depression" also correlates with the seasons. And the seasons may even be the causation.

Or it could be that people becoming depressed causes rain

This is really insteresting... http://www.google.com/trends?q=suicide%2C+depression%2C+god&ctab=0&geo=all&date=all&sort=0

God seems to follow a similar path but is more popular and there is a big bump in 2007 that doesn't correlate.

Can anybody find any other things that correlate to suicide, depression, and stress? http://www.google.com/trends?q=suicide%2C+depression%2C+stress&ctab=0&geo=all&date=all&sort=0

WOW-- drugs and pot fit almost perfectly

Drugs http://www.google.com/trends?q=suicide%2C+depression%2C+stress%2C+drugs&ctab=0&geo=all&date=all&sort=0

Pot http://www.google.com/trends?q=suicide%2C+depression%2C+stress%2C+pot&ctab=0&geo=all&date=all&sort=0

I've found that some things correlate at certain times. Look at Ugly from 2007 an onwards, http://www.google.com/trends?q=suicide%2C+depression%2C+stress%2C+pot&ctab=0&geo=all&date=all&sort=0 sick in the first half of 2006, http://www.google.com/trends?q=suicide%2C+depression%2C+stress%2C+sick&ctab=0&geo=all&date=all&sort=0 and even global warming in the first half of 2007 http://www.google.com/trends?q=suicide%2C+depression%2C+stress%2C+global+warming&ctab=0&geo=all&date=all&sort=0

How can that be explained?

drinking seems to correlate but be less popular http://www.google.com/trends?q=suicide%2C+depression%2C+stress%2C+drinking&ctab=0&geo=all&date=all&sort=0

[Moved to correct thread as requested – Tony]

And the slight decrease inb depression doesn't even indicate that it is actually dropping.
It might just be that people do not search it anymore (they commit suicide, they are more depressed and can't use a computer anymore...)

The possibilities I give are far-fetched, but I just want to highlight that you have to cross-check your data to draw any conclusion.
You'd need more data. If I remeber well, you have to use the Cramer V formula.

I was by no means drawing any conclusions. I was only pointing out the interesting correlations.

Many keywords seems to be decreasing slowly. Have you noticed this trend. It seems quite odd

I love how all these seasonal trends get that peak or dip around Christmas time.

Depression, drugs, rain, suicide and even God get that dip around Christmas time. Pot however sustains a high throughout the holiday season (yup, pun's intended)

also this (not a big surprise really) – http://www.google.com/trends?q=christmas%2C+toys&ctab=0&geo=all&date=all&sort=0

Keywords, that can somehow (if only intuitively) be linked to school, like [math, depression, rain, drugs, learning] corelate with the school holidays.

And I guess the happier people are, the more they cook: http://www.google.com/trends?q=depression%2C+cooking&ctab=0&geo=all&date=all&sort=0 . Or the other way around?

so this brings me to the question.. can google trends be used to predicate the future.. example. rain ::depression.. => leads to "some" school shooting type incident or similar disaster ??

if they right combo of terms used , can one predict an "incident" in the occurance ??

Google Time Machine: April Fool's 2008 :)

April 2008: Google Precrime
"Pre-organizing the world's information."

(That's a reference to "Minority Report".)

Here are the "turkey" and "diet" trends in select regions.

Canada shows "turkey" peaks in both October and November.
Canadian thanksgiving is in October.

http://www.google.com/trends?q=turkey%2C+diet&ctab=0&hl=en&geo=CA&geor=all&date=all&sort=0

In the region Turkey, the "turkey" searches do not seem to peak in November, but the news coverage does. Shown are all years and 2006.

http://www.google.com/trends?q=turkey%2C+diet&ctab=0&hl=en&geo=TR&date=all&sort=0
http://www.google.com/trends?q=turkey%2C+diet&ctab=0&hl=en&geo=TR&geor=all&date=2006&sort=0

In Russia, around May 2007, there seems to be a big jump, but not a peak.

http://www.google.com/trends?q=turkey%2C+diet&ctab=0&hl=en&geo=RU&geor=all&date=all&sort=0

Maybe this is a good tool for social and psychological research.

Regarding the diet searches: just look at probably the most common new year's resolution, i.e. 'give up smoking' http://www.google.com/trends?q=give+up+smoking&ctab=0&geo=all&date=all&sort=0

Staś, you can also see a spike on No Smoking Day in 2006 and 2007 too:

http://www.nosmokingday.org.uk/

Also, the number of searches in 2007 increase leading up to 1st July 2007 because smoking was banned in public places in the UK and Ireland. (In fact, if you select the region of United States you don't even get enough queries to draw a graph!)

You cannot see a correlation between rain and depression without having the percentage out of total searches. What if people use their computer less to search for during the chrismas holidays? Total searches for rain and depression will decrease, (the percentage of searches however should remain constant). Does this mean, that there is a correlation between those two words? Not necessarily.
I believe that peaks and downs in the graph reflect more the amount of total searches of that day rather than a strong dependency of the terms rain and depression.

Apparently American's only holiday is Christmas and normally they would go shopping ... kind of a perfect match:
http://google.com/trends/viz?q=holiday,+gift&date=all&geo=US&graph=weekly_img&ctab=0&sa=N
http://google.com/trends?q=holiday%2C+gift&ctab=0&geo=US&date=all&sort=0

Apparently Harvard is not as popular as Cambridge :P
http://google.com/trends/viz?q=harvard,+cambridge&date=all&geo=all&graph=weekly_img&ctab=0&sa=N
http://google.com/trends?q=harvard%2C+cambridge&ctab=0&geo=all&geor=all&date=all&sort=0

Why do people think that Google has a 's'?

http://www.google.com/trends?q=googles&ctab=0&geo=all&date=all&sort=0

Marekky and Clemens have got it. Most of these correlations are not meaningful. Put in anything you studied in school (physics, biology, to kill a mockingbird, etc.) and you see the same pattern: a sharp dip just before the end of the year and a long trough during the summer. Things that people are actually interested in, like porn and money do not exhibit this pattern.

Apparently Black Friday is growing in popularity over the years...
http://google.com/trends?q=black+friday&ctab=0&geo=US&geor=all&date=all&sort=0

I am a bit different, rain actually makes me feel happy

Throw eating into the stress, depression, suicide mix and you get
very unhappy overweight people.
http://www.google.com/trends?q=suicide%2C+depression%2C+stress%2C+eating&ctab=0&geo=all&date=all&sort=0

Gift and Savings:
http://www.google.com/trends?q=gift%2C+savings&ctab=0&hl=en&geo=all&date=all&sort=0
Apparently we only save for a few months before Xmas and then again at tax time.

Romance
http://www.google.com/trends?q=romance&ctab=0&geo=all&date=all&sort=0
I wonder was going on in late 2005 and late 2006?

Romance and Sex:
http://www.google.com/trends?q=romance%2Csex&ctab=0&geo=all&date=all&sort=0
Apparently whatever was going on wasn't as important as getting laid.

It's interesting that any comparison with "sex" gives you a graph of "sex" and a flat line at the bottom for the other term. I guess that biological drive is working really well.

I'd bet the differences between "suicide" and "depression" have something to do with suicide bombers in the news.

cars and beach have a nice positive correlation too – and opposite the school effect of "physics" or "math"

http://www.google.com/trends?q=car%2C+beach%2C+math&ctab=0&hl=en&geo=all&date=all&sort=0

There is an interesting thing going on with "mySpace" and "facebook" too

http://www.google.com/trends?q=myspace%2C+facebook&ctab=0&hl=en&geo=all&date=all&sort=0

^ add orkut to that http://www.google.com/trends?q=myspace%2C+facebook%2C+orkut&ctab=0&hl=en&geo=all&date=all&sort=0

Well, "depression" isn't an unambiguous word. It can also mean "low pressure area" and is specifically used in metereology, hence the connection with rain. And of course, correlation is not causation.

Apparently those in school google the word "sex" more when at home maybe??

http://google.com/trends?q=sex%2C+school&ctab=0&geo=all&date=all&sort=0

http://www.google.com/trends?q=tree%2C+snow&ctab=0&hl=en&geo=all&date=all&sort=0

Trees cause snow!

Oh, wait, what was that about causation?

> I believe that peaks and downs in the graph reflect more
> the amount of total searches of that day rather than a
> strong dependency of the terms rain and depression.

That's not the case. I asked Google whether Google Trends accomodates for general search traffic. Here's their answer:

<<Yes, Google Trends does accommodate for general search traffic. Therefore, an increase in search volume on Google Trends shows that the search term was searched for more relative to other queries.>>