Applying Machine Learning Technique for Knowledge Discovery in Network Database

Cyber attacks are malicious activities conducted in the digital realm with the intent of exploiting vulnerabilities, stealing sensitive information, disrupting operations, or causing damage to computer systems and networks. Network security is one of the viable ways of mitigating against cyber attacks. This study, through machine learning technique, was able to discover certain parameters that need to be taken cognizant of while working in a networking environment. An online network database retrieved from Kaggle was used in this study. Six inputs were used for the prediction of cyber attacks severity levels which was simulated with Naive Bayes algorithm in the Rapidminer Studio. The results show that the severity level “High” has the highest values for both raw and predicted data. It was also recorded that TCP has the least value (34%) for the predicted “High” severity level which shows that it is a good protocol to be used. On the other hand, HTTP had the highest value (67%) for the predicted “High” severity level which means that it is highly vulnerable to attack. With these results, internet users should make it of high priority to secure their data and network always by choosing the right protocols.