Flu Trends in a City chart from Google is Awesome!

Posted on January 15th, 2010 in Cool Infographics & Data Visualizations - 10 comments

There is nothing awesome about flu. In fact, it is anti-awesome if you have flu. I have been fortunate enough not to get it ever (it is another story that I have cold almost 364 days of an year).

Don’t worry if you are afraid of it, Folks at Google are not letting you down. They are using google search terms to predict flu trends and present the information in a beautiful chart.

Today I want to tell you why this particular chart on flu trends made by Google is awesome…

Take a look at it, I mean a really long serious look.
Flu Trends in a City chart from Google is Awesome!

Done looking? Good. Now I will tell you why it is awesome.

no, wait, I will tell you 5 reasons why it is awesome.
Flu Trends in a City chart from Google is Awesome!

  1. The chart lets you play: Even mission statements are exciting when you are playing with them (here is how). So when a chart lets you play with it, the chart instantly becomes awesome. In this chart, you can easily change the country or state or city to see flu trends for that. (related: making dynamic charts in excel)
  2. The chart highlights what is important: Look at the way Google shows current year vs. previous years. The current flu season (2009-2010) is shown in prominent blue color where as previous years are shown in dull shades of the same color. No more than 2 colors, but  still the message is loud and clear. (related: this year vs. last year charts)
  3. Beautiful Legend: The legend showing flu intensity is just beautiful. It follows the correct visualization principles (2 color gradient scale to show opposing ideas). Bonus awesome points for showing a tick mark on the legend for the current flu intensity.
  4. Correct chart: Google has access to almost any kind of data on flu trends, budgets and resources to make almost any type of chart. But they didn’t abuse that power to make an eye candy flashy “info-graphic”. Instead they made a line chart. Simple, time tested chart, and it just works. (related: excel line charts)
  5. Tells you when to worry: After all you are looking at the chart to know if you should be worried (and get a flu shot). So the chart tells that right away with the intensity heat map at bottom. You can take a look at current intensity of flu, compare it with previous year(s) and decide if you should get the shots for the little ones. Bonus awesome points for that, Google. (related: heatmaps in excel)

That is all.

I like this chart, so next week we will learn how to make this in Excel. Stay tuned… (and stay safe).

Related: Charting Principles, How to compare actual values with budgets

Reminder: today is the last day to vote for your favorite sales dashboard.

10 Responses to “Flu Trends in a City chart from Google is Awesome!”

  1. Loranga says:

    OT: I'm sorry to spam you once again with OT but I couldn't resist..
    At http://www.gapminder.org/videos/swine-flu-alert-news-death-ratio-tuberculosis/
    Hans Rosling shows a presentation and shows that the News/Death Ratio for Swine Flu is 8176 and for tubercolosis 0,1

  2. Chandoo says:

    @Loranga, my dear Swedish friend, Thanks for the link. Such a beautiful talk.

    I am sucker for Hans Roslings presentations and Gap minder in general. I am such a fanboy that I even built a gapminder like thing in excel... I have watched almost all of his presentations (ted talks) have linked to them here from time to time.

  3. Ken Puls says:

    Chandoo, another thing that I like about this is that, if I'm interpreting it correctly, the prior years are laid down in ever lightening colours. (So 2008's H1N1 effect in May/June 2008 is obvious.) That's kind of a cool idea.

  4. Chandoo says:

    @Ken.. yeah, the line thicknesses change too... pretty neat implementation (although it can be a bit distracting if someone watches closely...)

  5. Jeff Weir says:

    My thoughts:
    * Misleading representation of non-US data. Google published their finding in the journal Nature, where they say “we can accurately estimate the current level of weekly influenza activity in each region of the United States, with a reporting lag of about one day”. There’s nothing in that paper to say whether the correlation they observed between search phrases and flu indicators in the US would hold in different countries that had say a different level of internet access, or different ‘panic’ thresholds. So they should state this on this graph if users are selecting non-US countries

    *The legend is redundant. Beautiful or otherwise, it is effectively chart junk.

    * Would be good if some of the peaks from years past were labelled with the corresponding year. For instance, you see on the ‘National’ view that some time over the last 6 years there was a large peak in December. But you don't know which year it was unless you cycle through the years one by one. Why not label the highest point as ‘2006’.

    * You can’t compare directly between countries. Granted, it might not make sense comparing say Australia with the US if flu trends are largely seasonal (and they have adjusted the horizontal legend to fit each country’s flu season...i.e. for the US it runs from July to June, for Australia from Jan to December). But you might want to compare similar countries, and the data is certainly available. Here is where a ‘small multiples’ type series down the right hand side might be good.

    * They should make it more obvious that control are in fact controls. i.e. they should have ‘select year’ and ‘select country’ labels above the controls. And these controls should all be in the same place...as it is, two controls are on the graph, and two are in the left margin. Why?

    My overall thought: While this is a good start, is this really the best that Google can do?

  6. Chandoo says:

    @Jeff... Welcome back to commenting 🙂 We missed your insights...

    Very good points and very thorough analysis. Here is why I featured this chart,

    * it shows great restraint - nothing too fancy, but nothing boring either
    * the chart controls (atleast year select) is obvious. Anyone familiar with google interfaces (who isnt?) will notice the down arrow symbol. It is pervasive across various google sites.
    * I am not aware of the research behind this, but you have a valid point about search terms, their location and relevance to flu.
    * I also feel the same about labeling peaks. They should have added few labels to avoid guess work.

    that said, I really like this implementation and I want everyone to take a notice of it and see what good ideas we can pick up.

  7. Jeff Weir says:

    Hi Chandoo. Been on summer holidays, hence the lack of recent comments, plus between job contracts at the moment so most of my days are spent firing out CVs, and crying over salary offers.

    Don't get me wrong...this graph is praiseworthy to a point. The fact that its not a pie chart gets them one of your cyber-donuts rigth there. But I like to approach things from a "what could be made better" perpective. A graph isn't perfect until it is perfect.

    Your comment ..."Anyone familiar with google interfaces (who isnt?) will notice the down arrow symbol. It is pervasive across various google sites"...misses the point that the majority of people in the world have not yet used google for search, let alone come across any of their symbols on non search sites. I'm not sure that the majority of people even know what google is. Note that I'm saying the majority of PEOPLE, not the majority of COMPUTER USERS. Good design should make things as intuitive for new users as possible, because potential users will always outnumber actual users.

    I had to play around with things to see what was a control, and what wasnt. Sure, it took less than a minute to find out...but explicitly labelling the controls would have got that time down to less than a second. That's the kind of thing I expect from a world leader.

    Glad to be back. Keep up the posts.

  8. [...] week I have reviewed Google’s flu trends chart and told you why it is an awesome chart. This week, I am going to show you how such a chart can be [...]

  9. Gerald Higgins says:

    I also liked this alot, but there are also a few things that could be improved on that haven't been mentioned yet.

    1) Y axis scale.
    It LOOKS as if the Y axis scale has been set to automatically scale itself to whatever the highest level of intensity was FOR THAT LOCATION.
    And there is no indication of what exactly is meant by "Intense" at the top of the axis.
    This leads me to suspect that the actual level that matches with "Intense" probably varies across the locations, and this location-specific detail is therefore lost.
    For example, if San Francisco's peak intensity was 1 person in 1,000 infected, and New York's peak was 1 in 2, then these are very different scenarios which would look the same on the charts (all other factors being equal).
    Also, it prompts the question - even at peak intensity, should we be worried ? If peak intensity was 1 in 1,000,000, then maybe not.

    2) The UI includes a map of the USA, on which you can select cities.
    Some cities are difficult to select (for example in the San Francisco area).
    But more than that - the inclusion of a map creates the illusion that we are being given rich geographic data, when in fact we are not. The map itself is almost chart junk - a simple table would work better, by allowing easy selection of all locations. True, a map is helpful for selecting a city near your own location, if your exact location is not included, and for similar purposes.
    But the map does not allow real geographic interrogation.
    For example, this solution does not allow us to easily answer questions like -
    a) Which city got hit first ?
    b) Which city got hit worst ?
    c) Which cities pretty much escaped ?
    d) Did San Francisco get hit worse than New York ?
    e) How did the disease spread across the country ?

    3) I would suggest that comparing the exact timing of epidemic peaks from year to year is not of that much interest to the layman (although it may well be to doctors and scientists).
    A single chart showing the timing and height of intensity peaks for each city (or a selection of them) in 2009 would, I suggest, be more useful and interesting to most people than what Google have done.

    I'll say again, I liked this alot - overall I think this application presents a high standard that other providers of visual information would do well to try and emulate.

Leave a Reply