Network fraud number detection method and system, storage medium and terminal equipment

Abstract

The invention relates to a network fraud number detection method, and the method comprises the steps: firstly designing corresponding features for information fraud behaviors, carrying out the feature extraction of a user behavior log, constructing an original feature matrix, and carrying out the data preprocessing; then, according to the unbalance degree of the proportion of normal users and risk users in the original feature matrix after data preprocessing, carrying out minority class sample oversampling by adopting a self-adjusting oversampling algorithm, and reconstructing a training set; then carrying out pre-training, feature importance evaluation and feature screening through an XGBoost model; then, carrying out model training on the reconstructed feature matrix by using an XGBoost model and a LightGBM model; and finally, improving the model performance through a Stacking multi-model fusion mode, obtaining a two-layer model Logistic, and completing a mobile network risk user identification model. According to the invention, the accuracy and robustness of fraud number identification in network communication can be improved, and the actual application requirements are met.

Description

technical field [0001] The present application relates to the fields of machine learning and network security, in particular to a method, system, storage medium and terminal equipment for detecting network fraudulent numbers. Background technique [0002] With the continuous development of information technology and communication technology, network risk behaviors such as information and communication fraud are becoming more and more frequent, and technologies are becoming more and more advanced and diverse, seriously affecting people's life and work. Using big data and artificial intelligence to identify fraudulent numbers is an important direction to improve the technical ability to prevent and combat communication fraud. [0003] At present, the detection of fraudulent numbers is usually based on specific constraints, or using random forest algorithms. However, the rule-based detection method is not conducive to the changing situation of fraudulent behavior, and based on...

AI Technical Summary

Problems solved by technology

However, the rule-based detection method is not conducive to the changing situation of fraudulent behavior, and based on artificially designed specific rules, it is not flexible enough and the effect is relatively limited; the prediction accuracy of the random forest algorithm is low, so it is in the detection of fraud risk users. rate is relatively low

Images