java - Three arff files majority voting -

java - Three arff files majority voting -

i have 3 different arff files contain different classification info same instances, each line of each arff file concerns same instance, contains different info on instance. build new classifier have bulk vote on 3 classifiers applied on each arff info file cross validation

any clue or hint highly appreciated...

this basic proposition, based utilize java train , evaluate ensemble:

prepare each of 3 datasets according attribute requirements, or if possible, utilize same dataset 3 models using attribute filter each classifier (never tried this) train each of 3 classifiers using required training/attribute data code bulk vote rules @ end of evaluation process evaluate model on testing set.

there may other ways (such using vote combiner attributeselectedclassifier), doing code may give more command , flexibility trying combine.

java api machine-learning weka