Another econometrics resource that uses Julia: I am adapting my graduate level teaching notes so that the examples use Julia. These notes are at the first year Ph.D. level, and cover basic topics such as linear regression, up to more advanced topics such as simulation-based estimation and nonparametric methods. The notes are prepared using LyX, and are available in editable form. Export to LaTeX is straightforward.
The notes are at https://github.com/mcreel/Econometrics. The Julia conversion has been completed for Ch. 11 through 14, so far. I expect that the conversion will be completed around October of this year.
There is a support package, Econometrics.jl, which has the main routines, utilities, etc.
I am working in a panel data econometrics package which currently in beta test stage. Working on the documentation and unit tests these days. Might be something to check out for a few of those sections even if unobserved effects models is not included in the notes. UEM.jl
You have an interesting approach to programming, very structured. For my lecture notes, this is more in depth that what I present. However, I think I can learn some programming techniques from your code. Thanks!
Just a note that progress has been good on converting the examples to Julia. It’s almost completed, the only remaining chapters to convert are nonparametric estimation and simulation-based estimation. I have been adding some new material, too, including a MCMC method of estimating quantile instrumental variables models.
The support package https://github.com/mcreel/Econometrics.jl includes fminunc and fmincon, which loosely emulate Matlab’s calling syntax, in an effort to appeal to people coming from that background. There’s also a pretty useful simulated annealing routine…
I have now integrated the econometrics notes repo and the Econometrics.jl repo into the single repo GitHub - mcreel/Econometrics: Econometrics lecture notes with examples using the Julia language. So, the teaching related examples are in an Examples directory, and the underlying econometrics code is in the src directory. I deleted my Econometrics.jl repo, to help to free up that name for a possible future general-purpose package.
The code is intended to be used for teaching/learning the econometric methods, and tries to make the connection between the blackboard and the code as direct as possible, rather than going for highest performance. With Julia, we can pretty much have that cake and eat it too, though.
Here’s a visual example of how the notes/code integration works:
These notes still being updated, in preparation for the next time I use them to teach. There is now a script to compile a system image with all the packages, and also the most commonly used functions. It is pretty amazing to see the effect! TTFP is a dead issue, at least for this sort of usage case. Thanks to the people behind PackageCompiler.jl!
I’m working on updating the notes at https://github.com/mcreel/Econometrics. The README.md file gives instructions for using the examples. There is a Step 10:
To speed things up a lot , type ìnclude("MakeSysimage.jl") from Julia, having started Julia with julia --proj in the directory where you installed the notes. This will compile all of the packages, and keep the compiled images for re-use. If you do this, when you start Julia in the future, use julia --proj -J JuliaSysimage.so ( caveat : I have only tried this using Linux. If you’re a Windows or OSX user, please open an issue if these instructions don’t work)
If there are any Windows or OSX users out there who are interested, I would appreciating hearing about how the system image precompilation instructions work out for you.
