Linux log anomaly behavior prediction method based on XGBoost
Ruini Wang · 2024
To solve the problem that abnormal behavior in Linux logs is difficult to identify effectively, this paper proposes a method for predicting abnormal behavior in Linux logs based on XGBoost algorithm. By analyzing the historical log data, the method uses the efficient gradient lifting decision tree mechanism of XGBoost algorithm, and realizes the efficient learning rate and regularization strategy through multiple rounds of iterative optimization, effectively coping with the data imbalance and improving the model generalization ability. Experiments show that this model can significantly improve the accuracy and real-time of anomaly detection, and effectively distinguish between normal and abnormal log behaviors.