Multi-view time sequence abnormal point integrated detection and visualization method

Abstract

The invention discloses a multi-view time sequence abnormal point integrated detection and visualization method, including the steps: constructing a data point abnormity scoring matrix based on data point abnormity scores calculated by an abnormity detection algorithm; and decomposing the data point abnormity scoring matrix according to the number of the set visual angles, and integrating the datapoint abnormity scores under the visual angles to form more accurate data point abnormity score values, so that various time sequence abnormity detection algorithms are effectively integrated, and the accuracy of the data point abnormity scores is improved.

Description

Technical field [0001] The invention belongs to the technical field of computer applications, and specifically relates to a multi-view time series abnormal point integrated detection and visualization method. Background technique [0002] Anomalous points are data objects that are significantly different from the expected objects. Time series are a series of values ​​of the same statistical indicator arranged in the order of their occurrence. Time series anomaly detection is to identify data objects that are significantly different from the expected objects in the time series. the process of. The abnormal points in the time series often hide important information, such as information representing equipment failures and fraud. With the widespread application of time series data, time series data abnormal point detection has become an important research content in the field of big data mining, with important theoretical and application value. Its results are widely used in engine ...

AI Technical Summary

Problems solved by technology

A small number of integrated models mainly rely on data or target specific outlier detection algorithms, and lack a general time series outlier integrated detection model that can effectively integrate multiple time series anomaly detection algorithms

In addition, because time series may have various shapes and structures, visualization methods are an effective way to understand time series outliers. However, the current visualization of time series outliers is mainly based on a single perspective, and it is impossible to realize time series under different perspectives. Multi-angle comprehensive display and understanding of abnormal points

Images