What is Anomaly Detection?

Introduction

One of the uses of machine learning is to analyze data points for anomalous events or rare items. Accurate AI detection leads to prompt troubleshooting, which can help to reduce costs and improve the overall user experience. Microsoft Bing, Microsoft Office, and Microsoft Azure all have a range of data on a variety of metrics that are monitored using the anomaly detectors.

The anomaly is a data point that differs significantly from other data points. This data point is then referred to as an anomaly or an outlier. For example, we have a dataset that contains the income (in USD) paid by the taxpayers. Below is an illustration showing the dataset. There are some anomalies identified as shown in the illustration: