Orange Forum • View topic - Calibrating Scores when > 2 class values

Calibrating Scores when > 2 class values

A place to ask questions about methods in Orange and how they are used and other general support.

Calibrating Scores when > 2 class values

Postby sreastman » Fri Mar 15, 2013 12:23

I wish to use Orange.evaluation.scoring.CA when the class has more than two possible values. The documentation says "Many of these scores can be computed solely from the confusion matrix constructed manually with the confusion_matrices function." It goes on to say "If class variable has more than two values, the index of the value to calculate the confusion matrix for should be passed as well." Also, the first parameter of Orange.evaluation.scoring.CA , test_results, is described as either "ExperimentResults or list of ConfusionMatrix. "

Is this saying that a confusion matrix is necessary to calculate Calibrating Scores when the number of class values is greater than two, or just that is an option and that CA can still be calculated directly from Experimental Results?

If the confusion matrix approach is required, I wonder if anyone could provide a code example, including use of indices?

Thanks--Steve

Re: Calibrating Scores when > 2 class values

Postby Ales » Fri Mar 15, 2013 13:51

sreastman wrote:or just that is an option and that CA can still be calculated directly from Experimental Results?
Yes the CA can be calculated from the ExperimentResults directly (in fact it is the preferred way).

Re: Calibrating Scores when > 2 class values

Postby sreastman » Fri Mar 15, 2013 14:00

Great!

Thanks--Steve


Return to Questions & Support