bayes for generating new data

[andyspeak]

I am estimating a tree inventory for a city. I have already measured the heights and identified the species for about 1000 trees in some areas of the city and I want to estimate the remaining species composition based on the collected data. There are about 150 different species. I have heights and locations for the unmeasured trees from using remote sensing techiniques but I just need to populate these unmeasured trees with species estimates, preferably taking into account the height to predict the species.
I would like to use some kind of probability based method for this such as Bayes. Is this possible?

So to give an unmeasured tree a species I need a method that looks at the height and uses the measured height and species relationship to predict the species.

 

[Miner]

This sounds like an application for K-means cluster analysis.

 

[hlsmith]

Is this a very crude example of what you would like to do: look at height to predict species. If that is the extent you could create cut-offs for heights and use the simple bayes theorem.
If you want to look at a bunch of variable at once, I would create a poly-nominal logistic regression model. Though, I am sure you could do this as well using bayes, but I am not overtly familiar with those approaches beyond Bayesian Networks.

 

[andyspeak]

Thanks. Will give those a try