Programming Visualizations of "usage" stats

[Imran]

You may have seen the recent thread where I showed pokestat.org.uk. Behind this is a larger intent to put statistical analysis in the hands of the average player (rather than just programmers) by means of providing visualizations. The usage graph demonstrated is a simple example of the power of this approach, I have many other ideas.

However, there is significant work to be done on the backend to get this going - so I was wondering if anyone was interested in helping out with the front end. It will generally involve using ajax to pull data from REST APIs provided by me in the back end, and using javascript libraries to render stuff on the front end. (You can already see examples of me doing that in the GitHub repository.) I'm very pro you making a lot of design decisions (prefer Coffeescript to javascript?) and would like to make sure the thing works properly across all browsers (including mobiles) and is responsive (which it isn't at the moment!)

If you're interested in pitching in or want to know more, feel free to PM me to reply here. I'm happy to answer any questions :)

 

[novembermike]

I've been meaning to play around with D3 and this seems like a good excuse so I'd be interested in this. Tech wise I'd probably use Google's Web Starter Kit for a basic responsive design, React/Flux for a single page app, D3 for the visualizations and I'd probably use NodeJS and Express to render it (no offense but I'm not a huge Scala person and I like browserify, so Node wins for me on this). Since it'd be in a separate Github repo I'd also obviously make it MIT license. If that's all cool with you I can start playing around with it.

 

[novembermike]

Also, I just took a peek at your data and I have some recommendations. Right now this is the data you're returning:

{
"june": 18.09712,
"march": 14.75367,
"january": 14.19348,
"october": 17.17266,
"april": 16.92614,
"february": 14.49611,
"august": 16.51558,
"may": 18.0044
}

This will work fine until next January, when it will run into a duplicate key. Also, the data being passed across is relatively hard to use. If I want to print it out in a human readable form I need to write something to sort it and capitalize the first characters (and sorting it requires a transformation because january comes before june but after august if we're doing a string comparison). If the data is stored something like "YYYY-MM" then I can sort without transforming and then convert it to a human readable format relatively easily.