Abnormal data monitoring method and device, computer equipment and medium

Abstract

The embodiment of the invention is suitable for the technical field of machine learning, and provides an abnormal data monitoring method and device, computer equipment and a medium, and the method comprises the steps: determining a plurality of sub-sample groups from a preset sample group, and enabling the sub-sample groups to comprise positive samples and negative samples, the number of the positive samples in the subsample group is greater than the number of the negative samples; establishing a positive tree and a negative tree according to the subsample group; fusing the positive tree and the negative tree to obtain a combined tree; establishing a prediction model based on the plurality of combination trees in one-to-one correspondence with the plurality of sub-sample groups; and performing abnormal data monitoring by adopting the prediction model. The prediction model established through the method has high stability and effectiveness, and the accuracy of the monitoring result obtained through abnormal data monitoring based on the prediction model is higher.

Description

technical field [0001] The present application belongs to the technical field of machine learning, and in particular, relates to an abnormal data monitoring method, device, computer equipment and medium. Background technique [0002] When the data corresponding to a certain target is abnormal, it often means that the target has deviated from expectations, which may lead to mistakes or other losses. For example, when the account data of a bank customer is abnormal, it may indicate that the customer will go bankrupt, cause bad debts, and bring economic losses to the bank. [0003] By monitoring abnormal data, it is possible to predict targets with abnormal behavior, thereby reducing losses. However, the abnormal data in the data is often relatively small, and it is difficult to monitor the abnormal data through a general classification model or regression model. [0004] At present, when monitoring abnormal data, the entire user group can be grouped, and then different custo...

AI Technical Summary

Problems solved by technology

This method of abnormal data monitoring based on user portraits needs to re-execute the entire process including grouping and portraits as the data changes. The method is cumbersome and the system is redundant.

Moreover, the artificial division of customer groups is relatively random, and artificial portraits may also be unreasonable to a certain extent, resulting in inaccurate monitoring results

Images