Classification Algorithms Using Ensemble Methods

• Published in: International Conference on Future Internet of Things and Cloud (Online) pp. 168 - 175
• Main Authors: Ganiyu, Aishat, Darvishi, Iman, Addo-Quaye, Ronald, Yeboah-Ofori, Abel, Asare, Bismark Tei, Oguntoyinbo, Oluwole
• Format: Conference Proceeding
• Language: English
• Published: IEEE 19.08.2024
• Subjects: Artificial Intelligence; Bagging; Boosting; Bootstrapping; Classification; Classification algorithms; Decision Trees; Ensemble learning; Ensemble Methods; Industries; Machine Learning; Manufacturing; Medical services; Performance evaluation; Prediction algorithms; Predictive models; Python; Radio frequency; Random Forest; Regression
• ISSN: 2996-1017
• DOI: 10.1109/FiCloud62933.2024.00033

Machine Learning techniques are the backbone of successful Artificial Intelligence (AI) applications as they empower AI systems to make quality predictions and provide valuable insights from data that will aid decision-making within various industries. Ensemble method is a machine learning technique that helps to determine the suitable model for a dataset while limiting bias and variance; it is a technique used to obtain predictions by conducting maximum votes or averaging. This paper explores various Ensemble methods such as Bagging/Bootstrap Aggregation, Random Forest, and Boosting alongside their underlying algorithm Decision Tree for Regression and Classification cases. The contribution of this paper is threefold. The foremost objective is to perform a theoretical analysis of various Ensemble methods with sample pseudocodes to aid the implementation process. Secondly, we implement Ensemble method algorithms using basic Python packages such as numpy, pandas, math, matplotlib and dataset installation packages for performance evaluations. Finally, we applied the developed code to real-world datasets across various industries, including healthcare, manufacturing, and real estate. The results highlight the different parameters and their performance on the algorithms while observing the proposed Ensemble program for performance evaluation.