The progress of computer science has caused that many institutions collected huge amount of data, which analysis is impossible by human beings. Nowadays simple methods of data analysis are not sufficient for efficient management of an average enterprise, since for smart decisions the knowledge hidden in data is highly required. The great disadvantage of the aforementioned methods is that they “assume” that statistical properties of the discovered concept (which model is predicted) are being unchanged. In a real situation we could observe the so-called concept drift, therefore designing data mining methods, especially classification ones for data streams is currently the focus of intense research. On the other ha nd, we can usually use a number of classifiers for each of pattern recognition tasks which differ each other. Therefore developing combined classifiers has been mentioned as ones of the most promising trends in the pattern recognition which can exploit unique elementary classifier strengths and could adapt to the changes of classification models.