Data change identification network based on cost correlation and classifier point distribution method thereof

Abstract

The invention relates to the technology of enterprise data analysis and user power utilization behavior identification, belongs to the field of marketization load prediction and particularly relates to a data change identification network based on cost correlation and a classifier point distribution method thereof. Through enterprise data, each classifier of a data change identification network ina data stream in use is put on a data item, wherein the data item comprises original data, intermediate generation data and decision data; each data item is put on the classifier; the classier and the placement position of the classifier form a data change identification network; data directly obtained from an external or physical sensor is original data and is stored in numerous different original databases; on the basis of each flow link and intermediate decision requirements, an enterprise generates numerous intermediate generation data; and the decision layer of the enterprise carries outstatistical calculation to generate decision data. On the premise of guaranteeing identification accuracy, the point distribution cost of a data transaction identification network is minimized, an accurate guarantee is provided for enterprise detail data identification, and operation monitoring cost is lowered.

Description

technical field [0001] The invention relates to enterprise data analysis and user electricity behavior recognition technology, belongs to the field of market-oriented load forecasting, and in particular to a data change recognition network based on cost correlation and a classifier layout method thereof. Background technique [0002] The decision-making data of enterprise operation is obtained from numerous original data and intermediate generated data through complex statistical calculations to represent all aspects of enterprise operation status. Due to the development of machine learning technology, when business operation decision makers face changes in decision-making data, they can use machine learning models to judge the changes in relevant original data and intermediate generated data, so as to locate the root cause of decision-making data changes. At present, the commonly used method is to place a classifier directly at the position where the decision data is genera...

AI Technical Summary

Problems solved by technology

However, due to the huge amount of enterprise data, the judgment accuracy of only placing classifiers in the decision-making data layer is not high; at the same time, installing classifiers for each data item will increase the cost of judgment (each classifier needs to install a separate data Aggregation Server and Training Compute Cloud)

Images