Towards an improved classification model based on deep Learning and nearest rules strategy
Mohammed El Fouki, Noura Aknin, Kamal Eddine El Kadiri · Dialnet (Universidad de la Rioja) · 2019
In this paper we present a comparison between two improved approaches i.e. hybrid rules-based method with a proposed wrapper and nearest rule strategy, deep principal component analysis. We also perform several experiments with an analysis dataset from a distance learning platform. Several classifiers were developed to compare the performance of the proposed approaches, using accuracy, TP rate, F measure, PRC area, MCC, precision, recall and receiver operating characteristics area (AROC) as metrics. The result confirms the utility of these algorithms for classification and shows clearly the superiority of our approaches