Weka - differences between Explorer and Experimenter outcomes

问题

I just wondered why is the % correctly classified differs from the Explorer and Experimenter aspects of Weka. I have checked to ensure I am employing 10-cross fold validation as well as all other paramaters!

Anyone have any ideas?

Thanks

回答1:

I have the solution, as provided by Mark Hall, as I emailed him on the Weka Mail list. Here is the difference between Explorer and Experimenter:

The Experimenter operates differently from the Explorer. The Explorer sums evaluation metrics over the folds of the cross validation - e.g. percent correct is computed by summing all the correctly classified instances over the test folds and then dividing by the total number of instances. The Experimenter, on the other hand, computes averages over the folds. Furthermore, the default in the Experimenter is to run 10 repetitions of 10-fold cross-validation (so 100 folds are averaged over).

回答2:

Not sure but random seed might be different on Explorer and Experimenter. If this is the case, data sets will differ which results to different percentage.

来源：https://stackoverflow.com/questions/12495877/weka-differences-between-explorer-and-experimenter-outcomes