I want to develop a biomass model system including aboveground biomass model, belowground biomass model and stem, branch and foliage biomass models. Because the observations are different, so I deveolped two equation systems for aboveground biomass and belowground biomass respectively. But somebody suggest that we can use dummy variables for fitting both equation systems simultaneously. Could anyone give me detailed suggestion to solve the problem using dummy variables? Thank you.
yes, use dummy variables. you will need k-1 dummy variables where k is the number of different states you want to study.i see that in your case k=3 (above ground, below ground, and foliage. so you need two dummy variables. please see
https://www.youtube.com/watch?v=UhqCExVWw2Q
or, for an example of dummy coding you may download this paper.
http://papers.ssrn.com/sol3/papers.cfm?abstract_id=2630932
Dear Jamal, thank you. I will read the references you suggested at first.
Hello,
not sure what the point of dummy variables would be. I should think you could through all your equations and data into an optimization routine that finds the parameters for your models through numerical optimization. I guess, what that wold look like in detail would depend on your equations.
Best wishes,
Joerg
Hello,
I´m not sure than ussing dummy variables is the rigth way to fit simultaneously a biomass equation system, I have used a weighted restricted SUR fit () as there is correlation between the errors of the equations, with this approach you can get additivity among in your equation system. you can check the next post and publication.
best regards.
Jose
https://www.researchgate.net/post/How_can_I_set_a_system_of_equations_to_predict_biomass_using_weighted_restricted_SUR_in_SAS
http://www.srs.fs.usda.gov/pubs/ja/ja_parresol009.pdf?
Article Assessing Tree and Stand Biomass: A Review with Examples and...
Dear Jorg, Jose and Rogrio, many thanks for your suggeston an references. The additivity and heteroscedasticity of biomass models are not my problem. The real problem is that more than two equations are related to each other but the observatons for fitting each eqution are different, can we estimate simultaneously the equations as a system using dummy variables? For example, we have three equations:
(1) Y1=f(X) n=150
(2) Y2=f(X,Y1) n=150
(3) Y3=f(X,Y2) n=50
That is, for the variables (X,Y1,Y2,Y3), there are 150 observations for measuring X,Y1,Y2, but only 50 observatins for measuring X,Y1,Y2,Y3. This is the study case.
- Badges
- Science topic
- Similar topics
- Philosophy Of Science
- Models