With the exception of John Grinder all of the well known trainers in NLP teach Analytical Modeling they include Steve Andreas, Robert Dilts, David Gordon and Jonathan Altfeld. Each Modeling strategy has it's strengths and it's weakness and naturally some people are going to be better modelers with certain strategies than others.
Steve Andreas' style of modeling from my estimation seems to make great use of Contrastive Analysis he has written articles on Modeling and NLP Modeling which are on his website.
Robert Dilts has written an entire book on Modeling in which he stress people focus on features and pattern detection. A feature would be something like what metaprograms are being used, what body movements, submodalities, language patterns, etc... once features have been identified the modelers is to use Mill's methods in order to discern patterns.
David Gordon utilizes his experiential array in which all the features that one would search for are already given in a single easy to use format. They are simple to compare and contrast from multiple patterns in order to create their resultant model.
Jonathan Altfeld who came to the field of NLP from the field of Artificial Intelligence. He was able to map over some distinctions from his work in Knowledge Engineering along with some of Robert Dilts work in Sleight of Mouth in order to create a very interesting and useful method for Modeling people's thinking processes. It's a very interesting and useful method of modeling indeed.
Grinder's method which is by far my favorite from what I can tell or at least what I have attempted to do when I've worked to model people. What you do is identify your model. Have them engage in their competency that you are there to model as they do that you are to simply remain open and to absorb what they are doing until the point where you build up intuitions about what they are doing.
You then attempt to replicate what they are doing without any judgment other than did you replicate similar results in a similar context. Once you have done this you then identify an intuition without judgment once again eliminate it from your performance to see if it makes a difference if it does add it again and notice what occurs to make sure it is in deed essential to the process.
You continue through this subtraction process until you have cycled through all the pattern necessary in order to reproduce the models competence. Final step is coding of the patterning into a model.
Modeling is really an exciting skill. In my opinion the reason it isn't done more is that requires time and effort to accomplish. It's not always a project that can be completed in a short period of time. But it's worth it.