Evaluation of Algorithmic Metrics with A Focus on Server Cyber-Risks

Abstract

The purpose of the study is to evaluate algorithmic metrics focused on cyber risks of servers using: K nearest neighbors, Decision Trees, Naive Bayes and Logistic Regression (machine learning methods that apply classification algorithms), was conducted in order to select the best model to develop predictive standards to mitigate these types of threats. In addition, this research was conducted in order to help public and private entities in the field of cybersecurity to be able to have multiple decisions according to automation, security and process structuring of each algorithm. Another point is, the approach of the study is quantitative, based on a scientific research methodology. Therefore, the results in relation to the test scenario were obtained that each algorithm has demonstrated high performance both in operation and veracity in the face of multiple tests, but Naive Bayes has obtained better results for each metric, likewise, it was determined that such algorithmic parameters help in the analysis and processing of data in order to improve the accuracy of malware threats. In this sense, cybersecurity has been consolidated as a broad term that faces the challenge of strategically balancing multidisciplinary areas that require guidelines, spaces and priorities. As a main recommendation is to implement VPN services to manage the PCs used remotely by users because it maintains a secure data packet channel, they are not easy to breach, in addition, as specialists we must evaluate and validate severely the connection of the tools used by performing a scan for vulnerabilities and constantly employ technology that protects the entity's information.