Managing firewall systems to ensure network security while facilitating legitimate access poses a significant challenge due to the complexity of security rules and the risk of human error. This study addresses the need for effective firewall management by leveraging a powerful combination of the ELK (Elasticsearch, Logstash, Kibana) stack for data analysis and visualization, alongside machine learning models for predictive traffic routing. The paper aims to develop a model capable of accurately predicting firewall actions based on historical network traffic data, ensuring real-time decision-making, and providing explainability for administrators. Our findings demonstrate that machine learning models, particularly decision trees and convolutional neural networks (CNNs), can achieve high accuracy rates in predicting firewall actions, although tree-based models offer faster processing for applications requiring real-time responses. This paper presents a comprehensive approach that leverages the capabilities of the ELK stack and machine learning to build a more intelligent and adaptive firewall system, ultimately improving network security and performance.