Research on Network Intrusion Detection in Random Forest Based on Pearson Feature Selection
Xin Li · 2025
With the continuous escalation of cyber-attacks, improving the accuracy of network intrusion detection systems has become crucial. In this study, a new Random Forest algorithm—the Random Forest algorithm based on Pearson feature selection is proposed. First, feature screening using the Pearson Correlation Coefficient was performed to select the factors with strong correlation with being attacked and remove irrelevant factors; then, the remaining variables were modelled to form the improved Random Forest for the detection model of network intrusion, and finally, the improved Random Forest was compared and analyzed with other models. The experimental results show that the improved Random Forest model proposed in this paper achieves an accuracy of 90.16% on the test set, which is an improvement of 2.08%, 9.09%, and 20.17% over ordinary Random Forest, SVM and Logistic Regression, respectively. Simultaneously, the model's accuracy fluctuated by only 0.9% over 10 experiments, demonstrating a very high degree of stability. The model can quickly focus on core threat indicators, improve detection efficiency and accuracy, and provide ideas and technical references for network security practitioners to improve their network security protection.