[Gretl-users] Suitable use of libgretl?

Hello gretl-users, I'm looking to implement an application, and am wondering how best to go about it. libgretl looks like it might be the answer, but I'm not a statistics/machine-learning guy and would like a nudge in the right direction. I need to take multiple streams of data (integers) in real time for which dependencies are not known beforehand. From the data streams I need to infer dependencies and identify patterns, and mark data which do not match the patterns with a high confidence. As an example, stream A could be related to stream B by some function, linear or otherwise. If B has a good history of increasing/decreasing proportional to A, but then one day A decreases and B increases (diverging from the established pattern), that data needs to be marked. I'm intrigued to know if gretl 1) does this out of the box (I suspect I'd be too lucky) or 2) can help implement this thing, or 3) is completely in the wrong domain for this kind of application. Many thanks, Matthew