Transaction index abnormality monitoring method based on deep learning model LSTM

Abstract

The invention discloses a transaction index abnormality monitoring method based on a deep learning model LSTM. The method comprises the steps that 1 transaction index historical data are collected and processed; 2 the LSTM model with transaction index time series pre-determining is trained for the historical data processed in the step 1; 3 through the LSTM model of the step 2, the normal index interval floating value is calculated; and 4 the current index value is predicted through the LSTM model of the step 2, and abnormality capturing is carried out on the actual index value according to the normal index interval floating value of the step 3. According to the invention, the time series index is predicted through the deep learning model LSTM; the predicted current index value is accurate; the pre-determining accuracy of irregular time series indexes is greatly improved; subsequent abnormality capture is accurate; the output result predicted by the LSTM model is combined, and a residual fitting formula and a logistic regression model algorithm are used to calculate the normal interval floating value; abnormality capture is accurate; and the efficiency is improved.

Description

technical field [0001] The invention relates to the field of monitoring methods for abnormal transaction indicators, in particular to the field of monitoring methods for abnormal transaction indicators based on a deep learning model LSTM. Background technique [0002] In today's social enterprises, the degree of informatization is getting higher and higher, and the application of big data is becoming more and more extensive. More and more information of enterprises can be reflected through data, which is often called indicators; among them, transaction indicators are of great concern to financial enterprises. Whether it is a system failure or an external event, if it has an impact on the business, it is best to show it on the transaction indicators first; abnormal monitoring of the transaction indicators can detect and respond to problems in real time, improve business stability, and avoid unnecessary Loss. [0003] The existing technology generally completes the monitorin...

AI Technical Summary

Problems solved by technology

[0013] The purpose of the present invention is to solve the problem that the existing detection method abnormality capture is prone to errors in judging irregular timing indicators, and abnormality capture is heavily dependent on historical abnormality information records, and manual participation is required when there is no abnormality record, which easily leads to insufficient coverage of abnormality index values. The problem of comprehensively leading to inaccurate abnormal capture provides a more general, effective for regular and irregular time-series index abnormal capture, reducing manual participation in index abnormal judgment, efficient and accurate transaction index abnormality based on deep learning model LSTM monitoring method

Images