Voting Combinations-Based Ensemble: A Hybrid Approach
Abdul Ahad ABRO,Mir Sajjad Hussain TALPUR,Awais Khan JUMANİ,Waqas Ahmed SIDDIQUE,Erkan YAŞAR
DOI: https://doi.org/10.18466/cbayarfbe.1014724
2022-09-12
Celal Bayar Üniversitesi Fen Bilimleri Dergisi
Abstract:Machine learning (ML) is a prominent and extensively researched field in the artificial intelligence area which assists to strengthen the accomplishment of classification. In this study, the main idea is to provide the classification and analysis of ML and Ensemble Learning (EL) algorithms. To support this idea, six supervised ML algorithms, C4.5 (J48), K-Nearest Neighbor (KNN), Logistic Regression (LR), Support Vector Machine (SVM), Naive Bayes (NB) and One Rule (OneR) in addition the five UCI Datasets of ML Repository, are being applied that demonstrates the robustness and effectiveness of numerous approaches. In this paper, a voting-based ensemble classifier has been proposed along with two base learners (namely, Random Forest and Rotation Forest) to progress the performance. Whereas, for analytical procedures, significant parameters have been considered: Accuracy, Area under Curve (AUC), recall, precision, and F-measure values. Hence, the prime objective of this research is to obtain binary classification and efficiency by conducting the progress of ML and EL approaches. We present experimental outcomes that validate the effectiveness of our method to well-known competitive approaches. Image recognition and ML challenges, such as binary classification, can be solved using this method.
English Else