These guys do. I’ll explain who they are at the end of the blog. We will get there via a diversion about repository activity tracking and large data analysis.
New software being written
If you are ever curious about how much work is going on behind the scenes to bring you the POLARIS software, you can go to the github repository and check out some interesting graphs about this.
A good starting point is the Commits graph which shows how many software updates (“commits”) have been made to the code base each week, and with a particular week highlighted, the commits per day can be seen. Clearly Wednesday March 9th was a busy day…
(03/06 means the week starting Sunday March 6th, and the peak at 28 commits happened on the Wednesday 3 days later).
Still, on normal weeks we are still running at between 5 and 10 commits a week. This relates to new software, or corrections to existing software, being created on a daily basis, both by the main development team in the UK and by contributors from places like Denmark, Belgium and Turkey.
Another chart, which unfortunately is not publicly available like the one above, shows the number of different users looking at the repository on a daily basis, and when they visited. Here is the chart for the last week (and which arose my interest, hence this blog).
This two-week plot starts on a Wednesday and has two clear zero’s for the weekend of April 30th and May 1st. We have a peak on the following Thursday of 48 views and 8 visitors. Saturday May 7th is quiet as you might expect, but Sunday May 8th showed 18 views from 6 visitors.
Here is a clue to the mystery.
Bloomberg hosted the latest PyData conference in London from Friday 6th to Sunday 8th, which featured intermediate to advanced level presentations on the application of Python to solve Data Science challenges, from data management through analytics and visualisation. (OK – I copied that bit; I wasn’t there myself).
On the Sunday afternoon, a presentation entitled “Python flying at 40,000 feet” described some of the work being carried out to improve the quality of the information we derive from large volumes of flight data. If you are interested, see http://pydata.org/london2016/schedule/presentation/23/
The github tracking statistics show us that five people from the conference linked directly to our repository to follow up from that presentation.
So what about the photo at the top of this blog? The two presenters, Marko Vasiljevski and Raffaele Rainone, are first and third in the photo (counting from the left). A much more fun fact about this photo, which was taken at dinner before the conference, was that they are Croatian, Sri Lankan, Italian, Greek, British and Romanian respectively, and all work for Flight Data Services – but you’d guessed that already, I suspect.