New Economist writes: In a new post on the Statistical Modeling, Causal Inference, and Social Science blog, Aleks Jakulin at Columbia University points us to a great online tool, ZunZun. It lets you use 2 and 3 dimensional 'Function Finders' to 'help determine the best curve fit for your data'."
On William Greene's site I found a neat data set (Data Tables :: Table F6.1) for estimating a Cobb-Douglas production function. ZunZun comes up with the following suggestion:
Contour Plot:
The R2 reaches an unrealistic 0.968 (0.94 for the Cobb-Douglas specification). Textbook data... Here is a scatterplot of the logarithmized data (created with R):
On William Greene's site I found a neat data set (Data Tables :: Table F6.1) for estimating a Cobb-Douglas production function. ZunZun comes up with the following suggestion:
Y = β1( L0.5K0.5) + β2(cos(L)K1.5)
Surface Plot:Contour Plot:
The R2 reaches an unrealistic 0.968 (0.94 for the Cobb-Douglas specification). Textbook data... Here is a scatterplot of the logarithmized data (created with R):
Mahalanobis - am 2007-03-27 04:50 - Rubrik: mathstat
Hedgehog (guest) meinte am 27. Mar, 06:48:
We shouldn't take this shellgame too seriously...
It's a nice toy but, without checks and balances of solid (or at least plausible) economic theory, it just remains what it is -- a data mining exercise. I love cos(L) though; it makes your imagination run wild! :)
AleksJ meinte am 27. Mar, 20:59:
Once the author of ZunZun implements complexity control, such problems should become more rare. But clearly, you shouldn't pick the top ranked model alone, but instead one that makes sense. It's a tool, not artificial intelligence. You can overfit with linear regression too.
James R. Phillips (guest) antwortete am 28. Mar, 15:21:
Advice from zunzun.com's author
I suggest fitting the data set to a smoother function, for example a flat plane. Just my opinion. James Phillips
http://zunzun.com
