I ended up successfully completing two programs that Kari wanted.
I spent much of the last week fine-tuning those programs. The first, which I spent the majority of the last four weeks on, was focused on calculating the response rates of various groups to emails. I managed to write a neat little program that spit out an organized table displaying rates per group. It worked well and I even wrote a nice little interface to make it simple and accessible to non-programmers.
It is pretty useful in sending emails to know the response rate of certain demographics. For example, in a mass email sent out about Monsanto (a company that produces seeds for farming)- how many people who feel strongly about climate change opened that email? How many clicked through to the website? How many people gave money? How much? Having this data in hand allows campaigns to make more grounded strategic decisions, and can definitely give an edge both in raising funds and minimizing email costs.
However, the main problem with that program that I have been struggling to fix was that it was too slow. It requires that I download hundreds of thousands of data entries off of a database somewhere in the world. That process is slow because the company we work with did not provide fast ways to extract large amounts of data. It is not something I can easily fix, although Kari and I did spend two hours brainstorming solutions on Friday afternoon. I think if I had more time to work on it, I might be able to speed things up...
The second program focused on discovering how many members belonged to multiple groups, and what rates these groups had. For example, I looked at the cross-membership rates of how many people signed gun petitions versus how many people belonged to the vanilla democrat demographic. Not surprisingly, in that example, most of the people who signed the gun petition belonged to the democrat group, but not vice-versa. You can see how this can be useful in targeting analytics for emails- Kari can see which emails worked effectively for which groups, and re-target those emails to groups with similar membership compositions.
Like the previous program, the end result was that the program was too slow. Although I did manage to come up with a smart way of downloading the data and running it on a local machine, there is still a lot of work to be done.
There is still a lot of work to be done.
On Friday afternoon, after Kari and I spent two hours discussing future directions, he offered me an opportunity. Come back over the Summer, this time on the payroll, and try to finish things up. He said he is working on the actual terms such as pay rates and working hours, and a legitimate job description. Am I excited? Absolutely!