A Review of Anomaly Detection Algorithms in Times Series Data

Time series data has increased due to the increasing ability of systems to record and record data in small time units. In this regard, the need for methods to extract valuable information and analyze this type of data is a challenge. One of the important pieces of information that can be extracted is the discovery of existing anomalies. Anomaly means any type of event outside the normal process in a time series. Especially in critical systems, it should be detected in time to prevent problems from occurring. The presented methods are used in various fields, for example, hidden diseases can be detected in time, system problems can be investigated, and errors can be used. The methods presented for anomaly detection include clustering, classification, automatic coding, CUSUM-EWMA, etc. This article aims to study the recently available methods to detect anomalies and examine the strengths and weaknesses of these methods. Also, the application of each method and open issues in this field are also examined.

Keywords:

Anomaly detection , Time series , Valuable information , clustering and classification.