Time sequence anomaly detection labeling method and system

Abstract

The embodiment of the invention discloses a time sequence anomaly detection labeling method. The time sequence anomaly detection labeling method includes the steps that labeling information of labeled time sequence anomaly points input by multiple different labeling parties is acquired; anomaly result information is calculated according to the labeling weights and labeling information of all the labeling parties, wherein the anomaly result information is used for indicating whether all the labeled time sequence anomaly points are abnormal or not; after the labeling weights of all the labeling parties are adjusted according to the anomaly result information, the step of calculating the anomaly result information according to the labeling weights and the labeling information of all the labeling parties is returned to be executed, and the latest anomaly result information is output till two pieces of latest anomaly result information are consistent. By means of the time sequence anomaly detection labeling method, the technical problems that it is reflected that a time sequence deviates from a certain tendency only from numerical values in the prior art, and correlation with services is difficult are solved, and the accuracy of an anomaly labeling system is improved.

Description

technical field [0001] The invention relates to the field of computers, in particular to a time series anomaly detection and labeling method and system. Background technique [0002] Time series predictive analysis technology is based on ordered observational data sets associated with time sequence, using stochastic process theory and mathematical statistics methods to study the statistical laws obeyed by the data sets, so as to infer data development trends and guide the solution of practical problems. Scientifically and correctly predicting and analyzing various actual time series can produce huge economic and social benefits. Time series predictive analysis technology has been widely used in industry, address, ecology, economy, meteorology, medicine and other fields. [0003] In the existing time series anomaly detection, it is often based on the degree to which the time series values ​​deviate from the "normal" (assumed to be normal) sequence, and then use the posterior ...

AI Technical Summary

Problems solved by technology

However, the setting of the abnormal threshold only reflects the deviation of the time series from a certain trend numerically, and it is difficult to correlate with the business

Images