20 Mar 2007
I’m sure most of you have heard by now of Google’s acquisition of Gapminder’s Trendalyzer software.
Gapminder is a Stockholm-based foundation so it looks like the Trendalyzer development team will be moving to Mountain View with the foundation itself remaining in Sweden.
I’d never actually heard of Trendalyzer until now.
Over the past couple months, statistics and data visualization has become much more of a personal interest. From a personal perspective, Tendalyzer is a pretty cool tool independent of the potential business value it brings to a company like Google.
Having spent 5 minutes on Google’s site, it looks promising. It’s a given that they’ll be an integration with Google’s existing web analytics stuff but I’d love to see the basic trends in Google Reader improved as well.
Basically Trendalyzer provides a way of presenting data and statistics with simple time-lapse animation. What’s available now is mostly population data plotable in various *X vs. Y *forms (w/ a 3rd dimension being represented by color). For example, you could plot ‘Physicians per 1000 people’ vs. ‘Life expectancy, years’ and follow the progression over the last few decades for any number of countries. That’s just the beginning.
The ability to track multiple variables (countries and various other indicators) presented some interesting statistics. I certainly didn’t know that the life expectancy of Malawi (randomly chosen country) residents decreased by almost 5 years between 1993 and 2003 w/ the # of physicians per 1000 people also dropping from 0.026 to 0.011. Compare that with Canada’s constant 2.1 physicians per 1000 people (in 1993 and 2003) and life expectancy change of +1 years over the same time period.
Take India, since 1960 the life expectancy has increased from 44 years to 63 years. There’s been a 3x increase in physicians per capital over the same time period… 0.21 -> 0.6…
19 Mar 2007
I know it’s been done multiple times previously but I decided to play around with Google Maps a bit more.
My initial goal was to interface with our companies data in Salesforce and plot it accordingly using the Google Maps API. We have all sorts of conditions and variables attached to our data so it would (and did) make an interesting map.
*Problem: *It quickly became apparent that our salesforce account didn’t support API access. Evidently it cost more money. That being said, Salesforce does have a great API and developer community but without access to it, you’re pretty much S.O.L.
*Solution: *Even without programmatic access, I discoved that you can still create reports and export results to CSV. Nothing exciting but it did prove to be the easiest and most efficient way to get data out.
Once I had report data, I wrote a massager in python that took a CSV file, stripped out the useless information, and used Yahoo’s public geocoding service to translate addresses into appropriate latitude and longitude values.
The original plan was to build a django-based front end around it but that was scrapped in the interest of time. Instead I used a bit of php thrown together from various sources.
In the end it turned out to be a fairly straightforward (but interesting nonetheless) hack. I used a couple hours at the end of day to implement basic filtering capabilities. The data being displayed concerned sales territories, current and potential accounts as well as probability that any given account would convert. Different aspects were colour-coded and labeled.
At the end of the day I was quite happy with what I managed to produce. The Google Maps API is quite trivial to work with, the challenging part was finding worthwhile data to chart.
I’ll take some screenshots and post shortly.
13 Mar 2007
This coming Friday GenoLogics will be hosting its first Hack Day of 2007 (and second overall). As with the previous event, it’s internally focused with me serving as the organizer aka. the person who ends up doing the prize shopping.
Our company has been fairly receptive and has even committed to 4 hack days a year (roughly scheduled by quarters). There were some initial concerns over how these days would impact our 3 week development iterations but sanity eventually won out. If our schedule was so tight that we couldn’t spare a day, we’d be in pretty bad shape. It’d be nice to think that we’re not doing too badly.
The company hasn’t grown significantly over the past year (~50 people in total) so I would expect about ~20 people to participate.
Our last hack day occurred in November and was somewhat limited to developer (and the odd Product Manager) participation. This time around there’s been enough interest from other departments that the CEO has asked that everyone be included (not that anyone was being excluded previously, it was just marketed more towards coders).
The schedule is looking as follows:
8:00 – 8:30 : Start
12:00 – 1:00 : Lunch Break (pizza provided)
3:45 – 4:00 : Finish
4:00 – 5:00 : Demonstrations
You can’t help but notice there’s not a lot of time actually set aside for coding. If you’re intense enough you could always take the pizza back to your desk. The last hour has been set aside for walking demonstrations (where everyone gathers around each desk and the coder spends a few minutes explaining their hack). We didn’t even attempt to vote on hacks last time and likely won’t this time either. Everybody does an excellent job and should have an equal opportunity to take a prize home.
That’s about it. I’ll post some pictures and perhaps a movie next week.