Models for detecting and predicting anomalies in time series

This paper is devoted to the study of time series data representation, its construction, analysis, causes of anomalies, review of existing algorithms for finding anomalies, short-run anomaly prediction by methods such as Holt-Winters method, sliding window method, least squares method, exponential smoothing method, indicating the disadvantages and advantages of each method.

Autorentext

4th year students from the Department of Computer Technology and Information Security, Engineering Technology Academy, Southern Federal University.