I couldn’t resist participating in the recent O’Reilly webinar: Fitter, Happier: Improve Your Health and Productivity with R. There are many similarities between presenter Marc Garrett’s company, Intridea, and my own firm, OpenBI. We’re both in the app/dev-consulting space, Intridea in web development, OpenBI, analytics/big data. We’re similar in size, each in the 20-40 employee range. Perhaps most notably, both OpenBI and Intridea are distributed geographically, operating as virtual entities without central headquarters, tied together by customers and Internet technology. And both companies go to great lengths to assure a productive, collegial, happy work environment.
Add to that the focused use of open source software (Ruby, Rails, Pentaho, Python and Hadoop) as development platforms, a deep commitment to the usage of R for analytics, and an attraction for Strata Rx 2013, and I’d say we could be consulting cousins.
Garrett drives his concern for company productivity and employee well-being with a nifty data science app. Acknowledging his own plight of a “freshman twenty” weight gain at the beginning of his virtual work career, Garrett provides each of his employees a Fitbit step-tracking device to monitor/encourage physical activity and exercise. The company then has access to a social-facilitating Fitbit leaderboard website that tracks individual activity monitored by the devices to granular detail. For the 50% of employees that both use the device and allow Intridea access to their data, there’s a treasure trove of activity that can be monitored over time and related to company performance variables.
To get the Fitbit information, Intridea developed an API using their core Ruby on Rails expertise. The data are read and stored in JSON format in per-person folders on an authenticated file system. From JSON, it’s easy to produce files suitable for loading into the R Project for Statistical Computing for further wrangling and, ultimately, statistical graphics using the lattice and ggplot2 packages.
Garrett correctly notes that rather than divide the labor between Ruby-on-Rails and R, the entire development could have been completed in R. Indeed, much as I love Ruby, if I were the developer I’d have used Python and R, mainly because of the Python’s better suitability for scientific computing and interoperability with R. The issue’s moot though: either Ruby and R, Python and R, or R alone can handle the challenges.
If Intridea uses the Fidbit data as a proxy for employee well-being, they see commits on their Github version control site as a measure of employee productivity. The DS application clones repositories of the latest three months for subsequent processing in MongoDB. Intridea can then relate measures derived from the Github data to Fitbit metrics.
I can’t say I’m convinced on using Github data as a measure of firm productivity. It would seem that wily staff could game the code commit process to their advantage. The Fitbit data, on the other hand, can be used to construct time series measures of individual and company “fitness” that in turn relate to measures of company performance such as profitability and customer satisfaction. Both the Fitbit and the Github data are used to assess coverage of Intridea’s world-wide customer base by its distributed work team.
All in all, I really like what Intridea has set out to do, envisioning what’s been implemented so far as phase one of a “research” design that’ll be improved over time. I could readily imagine doing something similar for OpenBI’s RunKeeper data. The app well illustrates the potential of using data and analytics driven by freely-available open source tools to improve company performance.