I wouldn't mind taking a look at it if you can toss it my way.
Not a problem. PM me your e-mail and I will forward it to you.

In order to have a predictive analysis capability, you need a database and a very clearly defined model. One of the problems with all types of "class" analysis (actually, that is a sub-set of ascriptive group analysis and tends to be very reductionist) is that they tend to be very poor at the individual level, so figuring out "their influence on decision-makers" will be extremely tricky.
I've developed a rough draft model to illustrate the transformation from social context and relationships to decision-making. It's just might be good enough for government work! Of course, no model IMO can achieve 100% predictive analysis -- the real world is just too messy. The aim, though, is to provide a better contextual understanding of interests and motivations and how that manifests in decisions (in a COIN environment).

You talk about "relationships" between various groups, but how are you structuring your categorization of "relationships"? This is really tricky, since such categorizations often assume cultural norms that can render an analytic technique useless or, at least, less effective, outside of the boundaries of that culture (or culture area).
The model I am developing doesn't explicitly include categorization for class relationships; it explains the relationship between class and what I've termed 'masks' and 'factions'. Though I think a implicit component of the model is that class relationships are systems of patronage. The extent and shape of the duties and priveleges are dependent upon the details of a specific society; and what classes exist in the first place, I would argue in extension, are dependent upon that society's 'hub of power'.