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.