> Hi all,
> I have a simple learning problem ("simple" in the sense that it is easy
> to understand!).
> Consider the following website of house plans:
> http://www.planhouse.com/default.aspx >
> You will notice that the plans are categorized into sets. For any single
> plan, there is a list of features. For example, check out the "plan
> details" (in green) near the middle of this page:
> http://www.planhouse.com/plan.aspx?id=2361 >
> Every plan comes with a set of details like that.
> Now, I wanted to see whether I could use Weka to correctly classify
> plans. So, I copied down 30 sets of plan details from that website: 15
> in one category and 15 in another. I then used the Bayesian Network
> algorithm to attempt to classify them. No luck. It gave me about a 50%
> success rate (no better than random).
> The reason I chose Bayesian learning is simply because I won't have a
> large dataset. Now, I will increase the size of my dataset to see if
> that makes a difference, but I was wondering if anyone out there has any
> suggestions as to what I should be looking at?
I'd try a few other learning methods (such as logistic regression and
tree learners). If there is no improvement then it is likely that there
is very little relationship between your features and the category.