Order-preserving sequence rule mining method

Abstract

The invention relates to an order-preserving sequence rule mining method, which comprises the following steps of: obtaining all frequent order-preserving sequence modes corresponding to a time sequence, forming a frequent order-preserving sequence mode set, setting a minimum confidence coefficient threshold value minconf, and calculating a prefix sub-mode x of a frequent order-preserving sequence mode y, if the relative sequence of the prefix sub-modes is a frequent order-preserving sequence mode, obtaining an order-preserving sequence rule x = gt; y, iterating the process to obtain all order-preserving sequence rules; according to conf (x = gt; y) = sup (y) / sup (x), and x = gt is calculated; and finally, adding the order-preserving sequence rules of which the confidence is greater than or equal to a set minimum confidence threshold value minconf into the rule set R, calling the order-preserving sequence rules of which the confidence is less than minconf as strong order-preserving sequence rules, and calling the excavation of all the strong order-preserving sequence rules as order-preserving sequence rule excavation. According to the method, efficient frequent order-preserving sequence pattern mining is realized, then order-preserving sequence rule mining is performed on the frequent order-preserving sequence pattern, and an implicit relationship between the patterns is searched.

Description

technical field [0001] The technical solution of the present invention relates to the technical field of electrical digital data processing, in particular to a method for mining order-preserving sequence rules. Background technique [0002] Today is an era of big data, and many new problems have arisen from it. Many scholars have studied big data from multiple perspectives. The core of this research is to mine valuable information from a large amount of data, that is, data mining. Nowadays, data mining has been widely used in biomedicine, financial market, Internet and many other fields. As a very important research topic in the field of data mining, sequential pattern mining has received extensive attention for a long time. In order to solve various problems, sequential pattern mining has derived a variety of mining methods, such as negative sequential pattern mining can avoid the loss of frequent but missing items, contrastive sequential pattern mining can improve the cla...

AI Technical Summary

Problems solved by technology

However, this method ignores the continuity of the time series and cannot accurately mine the fluctuations of the sequence; the document "Order-preserving matching." published by Kim et al. proposed a representation method based on the order relationship, defining the relationship between elements as Binary relation (), but this document researches the pattern matching problem, needs to have prior knowledge; CN111581262A discloses a kind of order-preserving sequential pattern mining method, this method adopts the expression method based on order, in candidate pattern generation The pattern fusion strategy is adopted to realize the mining of time series trends, but this method only mines frequently occurring trends and ignores the implicit relationship between patterns. At the same time, this method uses the method of sequence-preserving pattern matching to calculate the support of patterns. degree, the relationship between subsequences is not used, so the original sequence needs to be scanned multiple times

[0011] In short, the existing research on time series mining has certain limitations, the mining efficiency is not high, and the mining results cannot be well utilized. It is difficult to ensure that important information will not be lost when processing time series, and so far There is not yet a good way to solve this kind of problem

Images