Best programming language for teaching econometrics?

Question

I will teach econometrics to learn in the fall. I think it’s important for them to learn programming using data as an important applied research. What suggestions do you have for a programming language. I tend mostly to R. What should I consider others?

Answer 1

R is a very good choice. Go for it.

The amount of resources on the Internet continues to grow. One good slide set is provided by the UCLA Stat Consulting Center .

And as you at Econometrics, make sure you look at Grant Farnworth econometrics from R to CRAN; Applied econometrics with R book of Zeileis and Kleiber is also very good.

Answer 2

Python - specifically the numpy, pandas, and matplotlib modules, using the ipython block as an IDE.

I will nod to other comments suggesting R. There is no doubt that R has a much larger library of packages to choose from. But I find it stupid to try to learn the R syntax as the first programming language. Beginners simply will not receive it, and they will be disconnected. R syntax is clearly incompatible.

From my own experience, using only smidgeon Python, I was able to get what functions, classes and objects were, and only after that R made sense to me. Moreover, I found that the ipython laptop is second to none, making coding and evaluating results easy. This eliminated the need to move eyes from the source editor window to the shell window and vice versa, as in any other environment.

Answer 3

I prefer R, but other free options to consider: a combination of an octave with gnuplot (Octave is a free implementation of Matlab)
python with numpy, scipy and matplotlib

Answer 4

I am surprised that no one mentioned Excel. As Brian Ripley said (see Slide 7):

Let's not feel: the most widely used piece of statistics software is Excel.

Indeed, Excel is a great tool for adding columns of numbers. Having said that, if the analysis you are doing is more complex, you should definitely use the right programming language.

Of the three obvious data management languages ​​(R, MATLAB, and Python), R has the best data management tools. See this other SO question for a more detailed comparison.

---

EDIT: While re-reading this, I am voicing a pretty pro-excel. I would like to expand my answer in order to maintain my reputation.

Excel causes me more problems than benefits. Its widespread use in my organization is mostly harmful. It is very difficult to keep track of where the data comes from and how your calculations work. Debugging Excel models is almost impossible. It encourages local data warehouses instead of central databases. It does not work with diff tools, and this makes reproducibility of your science difficult. From a semantic point of view, it does not separate data and what-is-done-into-data. The idea that all your variables need a location is distracting from understanding. The plotting possibilities are ridiculously terrible.

All that said, Excel is useful for a few specific things:

• As a CSV viewer. Of course, R has a View function, but it's not so pretty.

• Really simple data mining. Sorting, filtering, grouping columns. I find that this can be done a little faster using the point and click interface than with the code. Of course, you will have to write code later for reproducibility, but in the initial stages of Excel, it’s not bad for this.

• Charts are distinctive and easily distinguishable. If you see someone giving a presentation with a graph drawn in Excel, you cannot trust the results.

What is it. Anything else is a mess.

Answer 5

Since you are interested in R, you can also take a look at Incanter . Since it is built with the Clojure Dialog - Lisp for the JVM - you can use the vast array of existing Java libraries.

Answer 6

R Language is the best option for teaching and teaching an econometric course. Gretl is also another useful GUI software.