Update – 25/03/2009

Ok, so so far this week progress has been pretty good. The combining of the two techniques took a little longer than I first thought so I’m a little bit behind again. Although that is changing fast.

I was experiencing a problem in splitting the list of examples up so that only relevant examples were passed down the tree, but this was solved and now works perfectly. This was the part that took the longest and now this is done I can make some proper headway.

With this problem now solved, it was only a short step into building the tree properly. This step was almost seamless. It took a little over an hour and worked first time, meaning I finally have a decision tree built on past examples using the ID3 algorithm!! YAY!! The most annoying part of this process so far has definitely been that the most time consuming parts to code as the “little” things. What I mean by this is that most of my time has been spent fighting with entropy and lists of examples, but actually building the damn tree took all of 5 seconds by comparison. Still I guess that’s programming though eh?

I’m hoping to have the function that navigates the tree finished tonight or early tomorrow afternoon. If this is the case and there are no complications then I’ll be able to crack on with putting these classes into my OGRE application to give life to the model that my artist (Fei Hoi) kindly put together for me. I’ll put up another post when I get the tree navigation stuff in, but until then….have a video of the animations I’ll be using in the main application.