11
$\begingroup$

This question might sound very broad, but here is what I am looking for. I know there are many excellent books about econometric methods, and many excellent expository articles about econometric techniques. There are even excellent reproducible examples of econometrics, as described in this CrossValidated question. In fact the examples in this question come very close to what I am looking for; the only thing missing in those examples is that they are only research reports, without any mention of how the results of the study fared in a real-world application.

What I am looking for are documented/reproducible examples of real-world applications of econometric theory that ideally have the following characteristics:

  1. They should be reproducible, i.e. contain detailed description of (and pointers to) data, econometric techniques, and code. Ideally the code would be in the R language.
  2. There should be detailed documentation showing that the technique succeeded in the real world, according to a well-quantified measure of success (e.g. "the technique helped increase revenues because it enabled improved forecasting of demand, and here are the numbers involved")

I am using the term econometric quite broadly here -- I mean any sort of data-mining, statistical data-analysis, predictiion, forecasting or machine-learning technique. One immediate problem in finding such examples: many successful applications of econometrics are done in a for-profit setting and are therefore proprietary, so if a technique worked well, it probably will not be published (this is especially true in the case of proprietary trading strategies), but I am nevertheless hoping there are published examples that have at least property (2) above if not both (1) and (2).

$\endgroup$
6
  • $\begingroup$ @pchalasani, Does LTCM count? ;) $\endgroup$
    – cardinal
    Feb 8 '11 at 3:00
  • $\begingroup$ What about discrete choice modeling? I'd think most people would consider DCM to be a branch of econometrics and the entire travel demand forecasting industry relies on DCM to generate inputs for their forecasting models...that may not be what you had in mind though. $\endgroup$
    – Chase
    Feb 8 '11 at 3:06
  • $\begingroup$ @pchalasani, yes, how broadly are you interpreting the word "econometric"? Does, for example, Netflix's "Cinematch" system count (or any of the other successful "recommender" engines out there)? $\endgroup$
    – cardinal
    Feb 8 '11 at 3:23
  • $\begingroup$ Arnold Zellner isn't a bad name to look up. larry bretthorst's phd thesis on spectral estimation is also good $\endgroup$ Feb 8 '11 at 7:36
  • $\begingroup$ @chase and @cardinal: yes Travel demand forecasting and Netflix Cinematch ( and other recommendation engines) would be great examples, if there were any documentation about how they work, or at least a description of the underlying techniques. Other examples could be from non-profit domains such as education or government. $\endgroup$
    – R_Coholic
    Feb 8 '11 at 11:47
4
$\begingroup$

As mentioned in the comments, travel demand forecasts often use inputs from discrete choice models (multinomial logit, nested logit, mixed logit, etc) to assist in the development of behavioral mode choice or route assignment in their travel demand forecasts. DCM certainly has many applications outside travel demand forecasting, but it has been used in the transportation industry for 30+ years so there should be lots of good examples.

As for reproducible examples:

  • Biogeme is an open source piece of software that is optimized for estimating logit models. The website provides the data, code, and a paper write up describing their methods.
  • travelR is a project to make travel demand forecasts with R. There was a presentation at useR! 2010 about the project, abstract here and slides here. There is also a webinar coming up in a few weeks regarding R and travel demand forecasting that I'll find the link to and update here for those interested.
  • Transportation Review Board Conference has made a list of all papers available online this year. I don't have a specific paper to link to, but there are several committee's worth of papers regarding the application of choice models in the transportation context.
$\endgroup$
2
  • $\begingroup$ thanks for the details @Chase, DCM is an area I'm not familiar with, but looks interesting. $\endgroup$
    – R_Coholic
    Feb 8 '11 at 16:43
  • $\begingroup$ @pchalasani - for a more general overview of discrete choice modeling, Kenneth Train's book is a very good place to start for both theory and practical applications. It's a fairly dense read, but well worth the effort if you are interested. $\endgroup$
    – Chase
    Feb 8 '11 at 16:58

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.