Performance Evaluation and Comparative Study of Machine Learning Techniques on UCI Datasets and Microarray Datasets

Classification techniques are a very effective way to classify the data which is essential in the decision-making process. In the previous literature, several classification algorithms have been used in various applications such as biomedical, security, text classification, and image classification. However, the classification accuracy falls under some limitations due to the imbalanced data. This study has used five widely known machine learning techniques: Naive Bayes, artificial neural network, decision tree, k-nearest-neighbor, and support vector machine on four UCI datasets and one micro-array dataset. This study mainly concentrates on the functionality and the Advantages and Disadvantages of each technique. Some metrics have been employed to assess their success, including accuracy, precision, recall, F _score, and Matthew's correlation coefficient(MCC). The datasets are used in this study to highlight the evaluation of numerous metrics of each classifier, demonstrating that no single indicator can convey all information about a classifier's performance and that no single classifier can satisfy all classification requirements.