Flu Trends Timeline

This past week I came across this CSV-formatted data from Google. It’s search engine trends for searches about the flu for different parts of the US.

I thought it would be interesting to display this data on a map with a slider along the bottom to select the date. Here’s the result of spending two hours doing just that. I’m really happy with how it turned out, and, further, I picked up a few new tricks from the process.

You probably noticed there was a spinner when you first opened the page. This is because it’s asynchronously fetching the latest data.txt from Google. However, since it’s a cross-origin request, and I don’t control the server headers (static hosting), it’s using AnyOrigin.com to translate it into a JSONP request. That’s a really handy service!

To parse the CSV format, I’m using jquery-csv. I wouldn’t mention it except that it has a really cool feature I haven’t seen in any other CSV parser: instead of reading the text into a two-dimensional array — which would need to be “parsed” further — it can read in each row as a object, using the CSV header line as the properties. This is the toObjects() function. It makes it feel like reading straightforward JSON. For example,


Will be parsed into this in JavaScript structure,

[{name:"apple",  color:"red",    weight:"1.2"},
 {name:"banana", color:"yellow", weight:"1.6"},
 {name:"orange", color:"orange", weight:"0.9"}]

With the flu data, it means each returned object is a single date snapshot, just what I need. The only data-massaging I had to do was mapping over each object to convert the date string into a proper Date object.

Using two neat tricks I’ve got the latest data parsed into my desired data structure. Next up is displaying a map. At first I wasn’t sure how I to do this cleanly but then I remembered an old DailyProgrammer problem: #89, Coloring the United States of America. SVG maps tend to contain metadata describing what shape is what. In this case, each state shape’s id attribute has the two-letter state code. Even more, SVG plays very, very well with JavaScript. It can be manipulated as part of the DOM, using the same API, including jQuery. It also uses CSS for styling.

The tricky part is actually accessing the SVG’s document root. To do this, it can’t be included as an img tag. Otherwise it’s an opaque raster image as far as JavaScript is concerned. It either needs to be embedded into the HTML — a dirty mix of languages that should be avoided — or accessed through an asynchronous request. Accessing remote XML was the original purpose of asynchronous browser requests, after all (i.e. the poorly-named XMLHttpRequest object). I can host this SVG from my own server, so this isn’t an issue like the CSV data.

HTML doesn’t have a slider input, unfortunately, so for the slider I’m using the jQuery UI Slider. I’m not terribly impressed with it but it gets the job done. Even before I had the slider connected, I could change the display date on the fly from Emacs using Skewer.

In regard my initial expectations, this project was surprisingly very well suited for HTML and JavaScript. Being able to manipulate SVG on the fly is really powerful and I doubt there’s an easier platform on which to do it than the browser.

blog comments powered by Disqus

null program

Chris Wellons