Comparative Study of the Ensemble Learning Methods for Classification of Animals in the Zoo

Abstract

Understanding that Machine Learning concepts can help generate a strong classification system and provide a good prediction on out-of-sample data which allows us to realize it’s importance in fields of Taxonomic Classification and makes it easier to handle huge datasets. In this paper, we are aiming to solve a classification problem using supervised machine learning in Python. The zoo dataset taken from UCI data repository contains data items that describe animals according to certain attributes that categorize them under seven different classes. Our central aim in this paper is to provide a detailed comparative study of few of the major ensemble learners with respect to the base learner. We are also looking into all the factors that are primarily responsible for reaching the conclusion and how they affect the decision and select the most important ones. Our results show bagged decision tree as the best performing ensemble classifier with an accuracy score of 96.94%.