Select weight by accuracy

fiddinyusfida

Hi everyone,

I am conducting a classification experiment by using weight. My scenario is I divided the whole dataset using 10-Fold Cross-validation. In every training phase, I want to obtain some weights by accuracy results. And the final is applying the weight into the testing phase.

For example, I have 3 classifiers, the result from these classifiers are SVM = 80, NB = 60, and RF = 90.

Is it possible to obtain weight from the specific measurement like accuracy during training phase?

Telcontar120

Can you clarify what you mean by "weights by accuracy"?
Are you trying to build some kind of voting ensemble where you weight the vote by the overall accuracy?
In either case, why would you ever want to look at this accuracy only from the training phase?  That's not usually sound since training accuracies are notoriously overfit.