Cumulative effect identification method combined with dynamic and static similarity analysis

Abstract

The present invention proposes a cumulative effect identification method combining dynamic and static similarity analysis, which is characterized in that for the qualitative and quantitative identification of cumulative effects in time series, when selecting similar samples, on the basis of selecting "static similar samples", At the same time, "dynamic similar samples" are selected, that is, to change and improve the original selection mode considering the similarity between samples at different times, and to combine the sample change process between a certain time and several previous moments, and a method similar to the time to be studied and several previous moments, through Compare and analyze the results of the static similarity method and the dynamic similarity method, and identify whether the cumulative effect exists according to the theoretical and calculation results. The present invention can effectively solve the problem of identifying the existence of the cumulative effect, and by perfecting the selection rules of similar samples, the change law of the obtained time series data is more in line with the actual situation, so as to improve the coverage of mining and the precision and accuracy of realizing the mining purpose .

Description

technical field [0001] The invention belongs to the technical field of data mining, and in particular relates to a cumulative effect recognition method combining dynamic and static similarity analysis. Background technique [0002] Many data in daily life appear in the form of time series. As a commonly used data type, time series objectively records the information of the observed system at various points in time. Related research methods have been obtained in the analysis of various data. widely used. In 2013, China officially opened the era of big data, and various information and data in each industry showed explosive growth. Facing the processing of massive data, data mining is a necessary choice. Data mining is a comprehensive discipline, which combines important theories of other basic disciplines. Its main function is to extract useful patterns and rules hidden in large amounts of data. Its core work is to analyze the interaction between various parameters. law. A...

AI Technical Summary

Problems solved by technology

[0005] In the data mining algorithm that does not consider the cumulative effect, the static similar algorithm is used for the processing of the time series, and the data similar to the target parameters and other influencing variables are generally mined from the historical data. This method does not consider the existence of data. Correlation, leading to large deviations in the results

Images