Improved Network Threat Detection Using Random Forest Algorithm with Django-Based Output Visualization

The emergence of artificial intelligence (AI) methods has transformed network security by facilitating the use of predictive modelling for threat detection. This abstract to present an innovative methodology to improve network security through predictive modelling using emerging AI methods. Predictive modelling using AI entails processing a large volume of data of network traffic that uses AI algorithms to identify patterns within large data volumes to determine the possibility of threat(s) such as malware or intrusions that may result in security violation. The predictive modelling as proposed in this methodology allows for predictive models to determine the vulnerabilities within the network, and can further allow for the predictive modelling to give early detection of threats before they escalate into security violations. The methodology allows network defenders to exercise defendable action to improve the resilience of networks to absorb threats and allow for the constructive feedback to improve network defense. Through the use of AI and predictive modelling, this proposal ultimately seeks to add to developing practical and resilient security and network defense in an increasingly interconnected digital landscape. In this project, the three different algorithms will be described: Bernoulli Naive Bayes, Adaboost, and Random Forest algorithm. In my previous research, I used Bernoulli Naive Bayes and Adaboost algorithm, with the accuracy for Bernoulli Naive Bayes algorithm being 42% and Adaboost algorithm being 78%. In this project, I used the Random Forest algorithm, and the Random Forest Algorithm was executed on the project and reported 98%.