Time sequence clustering method and system, equipment and medium

Abstract

The invention discloses a time sequence clustering method. The method comprises the following steps: acquiring a plurality of time sequences and extracting a feature vector corresponding to each time sequence; acquiring a clustering K value of the plurality of feature vectors; clustering the plurality of feature vectors by using a preset clustering algorithm and the clustering K value; and clustering the plurality of time sequences according to a clustering result. The invention further discloses a system, computer equipment and a readable storage medium. According to the scheme provided by the invention, the feature vector of the time sequence is extracted, the dimension reduction of the time sequence is realized, the data after the dimension reduction is clustered by using the clustering algorithm, and finally the clustering analysis of the time sequence is realized. In this way, it is guaranteed that the dimensions of the feature spaces of all the time sequences are consistent, meanwhile, the sampling step length of the time sequences is not required, and time information is also considered when the feature vectors are extracted.

Description

technical field [0001] The present invention relates to the field of clustering, in particular to a time series clustering method, system, device and storage medium. Background technique [0002] Time series is the most common type of data. At present, most of the time series analysis focuses on the prediction of time series. But for some problems, the morphological comparison of time series is also an important problem. For example, the daily average price of various commodities (or the daily closing price of stocks) constitutes a time series. How to evaluate the consistency of commodity price trends can be attributed to the time series shape clustering problem. [0003] In order to ensure the reliability and stability of the system and service, the monitoring system has gradually become an indispensable system for every company and enterprise. With the increasing number of services and machines, how to analyze massive time series KPIs has become the first problem we need...

AI Technical Summary

Problems solved by technology

[0004] Since the dimension of the time series (sampling time point) is generally relatively high, some even reach tens of thousands of dimensions; secondly, the time series changes with time, so time information is included in it, if it is simply carried out The calculation of the similarity ignores the time information; and the time steps of different time series are not necessarily the same, some samples are sampled every second, and some are sampled every minute, resulting in inconsistent dimension sizes

Images