A simple exploratory analysis of my coding stats. I use WakaTime since late 2016 to track my use of programming languages and periodically fetch the daily stats from the WakaTime API (see e.g. my coding-stats.json Gist). Since recording only started in 2016, large chunks of C#, C/C++, Matlab and other languages are missing.
In this repo, I ran some trivial analytics from those JSON files. If for whatever reason you are
interested, the redacted.csv contains the language-specific information. I excluded other source
data from this repo since it can hypothetically contain mildly sensitive details such as project
and machine names.
I mainly wanted to see how my Rust journey went compared to other languages and at the very least it's interesting to say that within the last two-ish years as of writing this, I spent as much time writing Rust code as I spent time writing Python code since the recording started.
The C# curve is pretty linear, which is not surprising since I have been working in .NET focused companies since 2008 and therefore use C# a lot during my work life. What is interesting, however is that for the last three years I have not actively assumed the role of a software developer and still maintained a linear progression there.
This also tells a story about COVID. During the period of 2020..2022, a lot of work was done from home. The below plot shows the deviation of my actual C# coding times from a linear regression model (R²≈0.99) in terms of cumulative sum of programming hours. As is obvious, this number rises both in late 2019 and late 2020, to slowly and then drastically decline in 2022. If one were to look for symptoms of a burnout, they're right there.
There is also an interesting apparent seasonality in the deviations that would be interesting to analyze further.