A purpose for data mining is capacity planning. we would like to approximate parameters
and be able to identify (based on current trends) what the result might have been.
Does Mondrian have such a capability (or at least a way to plugin such a capability)?
Mondrian has the ability to do what-if scenarios. However, since it's an engine, that has to be exposed via a tool. Analyzer (part of Pentaho EE) does not currently support this capability. Saiku, however, does. I'd go check them out http://dev.analytical-labs.com.
Chapter 11 covers it, but briefly. It does cover the general context of question you're asking - based on what I'm seeing today, and another assumption Y, what does Z look like next month. It's not trivial (this type of stuff rarely is) but it's supported in Mondrian and Saiku (not Pentaho).
There's also some ability for Mondrian, speaking MDX to do linear regressions to forecast trend lines (smoothed) in general.