A new method of building a more effective ensemble classifiers

In this contribution the new method for selecting classifiers to build an ensemble classifier is presented. Sometimes we get many classifiers that classify an object based on various premises (attributes). Many of them are of low quality, therefore a new classifier is being built that takes into account the weight of individual classifiers. In general, it gives better results than individual classifiers. However, the use of all classifiers (especially those of low quality) does not always give satisfactory results. Therefore, we present a method that allows to eliminate some classifiers while increasing the quality of classification.