Analyzing the Performance of Anomaly Detection Algorithms

Abstract

An outlier is a data observation that is considerably irregular from the rest of the dataset. The outlier present in the dataset may cause the integrity of the dataset. Implementing machine learning techniques in various real-world applications and applying those techniques to the healthcare-related dataset will completely change the particular field’s present scenario. These applications can highlight the physiological data having anomalous behavior, which can ultimately lead to a fast and necessary response and help to gather more critical knowledge about the particular area. However, a broad amount of study is available about the performance of anomaly detection techniques applied to popular public datasets. But then again, have a minimal amount of analytical work on various supervised and unsupervised methods considering any physiological datasets. The breast cancer dataset is both a universal and numeric dataset. This paper utilized and analyzed four machine learning techniques and their capacity to distinguish anomalies in the breast cancer dataset.