From wikipedia:

"Heuristic (pronounced /hjʉˈrɪstɨk/, from the Greek "Εὑρίσκω" for "find" or "discover")

A heuristic method is used to come to a solution rapidly that is hoped to be close to the best possible answer, or 'optimal solution'.

A heuristic is a "rule of thumb", an educated guess, an intuitive judgment or simply common sense. A heuristic is a general way of solving a problem"

The purpose of this blog is to provide brief heuristics to help me (and perhaps others) understand topics in econometrics and quantitative methods more concretely.

In graduate school, I completed several courses in statistics and
quantitative methods including econometrics, mathematical statistics,
experimental design, mathematical economics, biometrics, and statistics
based courses in population genetics and plant breeding (my interest in
graduate school focused on the environmental and economic ramifications
of agricultural biotechnology). In my current employment, I'm constantly learning and applying new data mining algorithms and statistical techniques.

This blog is one way for me to quickly summarize and catalog the key elements of the techniques that I have used in the past, as well as new ones I encounter. While many of these concepts may transcend the range of topics normally thought of as 'econometrics,' most economists would find some of them very useful in their work. As a heuristic guide, some details may be compromised from time to time to illustrate essential themes. (just as with models from economics, sometimes it is necessary to abstract from ancillary details to better elucidate core concepts).

I often make use of SAS software or R code to illustrate many topics I find interesting in data mining and applied econometrics. I often find that coding allows me to get my hands dirty and forces me to understand with much more precision and greater detail exactly what's going on under the hood with regard to many of the statistical techniques and algorithms that often get encapsulated behind the GUI facade of point and click software.

I follow a 'recipe' for blogging very similar to stats blogger Jeremy Kun:

1) Identify a topic that sounds fascinating or something that I would like to master in greater detail

Or

Encounter a problem on the job that requires greater knowledge or detail of some technique I've never used or one that I'm familiar with but never applied professionally, or a topic that I would like to adopt for a classroom application in one of the courses I teach.

2) Research the literature related to the technique (often including journals such as the Journal of Applied Econometrics, Review of Economics and Statistics, Econometrica, The American Statistician, Journal of Applied Statistics, as well as numerous blogs and websites related to data mining and statistical programming)

4) Write a blog post that provides the theoretical background related to the topic of use and demonstrates its application in a simple way.

5) Update the post with related links and concepts, or new insights that I develop as I become more familiar with or apply the technique professionally.

Dear Dr Bogard,

ReplyDeleteI must say I find your Blog very illuminating. In the absence of an email, I decide to post my Question here; I have a Panel data of 30 variables for 10 years and across 19 sites. I want to run a panel VAR,but its proving difficult to do so, is it the okay to run a Fixed Effect regression on the panel instead of VAR? and then I want introduce the spatial dimension to it later and compare the two models (with and without space). Could you be kind enough to advise? Awaiting your favorable response. I can be contacted; wunduwamarum@gmail.com Thanks