Analysis of Cyber Risk Awareness Through Classification Algorithms
Session
Computer Science and Communication Engineering
Description
Exposure to cyber risks has increased significantly with the development of technology. Sophisticated attacks such as phishing, ransomware, malware, and credential misuse represent serious threats, while the limited awareness of technology users makes them highly vulnerable. This thesis develops along two parallel dimensions: assessing the level of cyber risk awareness among the citizens of Prizren and analyzing the potential offered by classification algorithms in preventing cyber-attacks. Data on citizens’ knowledge, practices, and perceptions of cybersecurity were collected through a questionnaire. The analysis of results revealed an acceptable level of awareness regarding the general concept of cyber risk; however, knowledge about specific threats was found to be limited. The thesis is divided into six chapters, including a literature review and methodology. In the technical part, the study examines the use of classification algorithms in ma-chine learning – including logistic regression, decision trees, random forest, and neural networks – with a special focus on XGBoost. This algorithm, known for its accuracy and flexibility, was used to automatically classify open-ended questionnaire responses, demonstrating its ability to categorize textual data into categories such as suggestions, issues, or lack of response. The results showed high accuracy and efficiency, making XGBoost a very reliable tool for practical and educational analyses. The final findings highlight the importance of integrating advanced classification algorithms into educational and protective systems. This thesis contributes to the scientific literature by combining empirical analysis with modern machine learning methods, providing a foundation for developing more effective strategies to tackle contemporary cyber challenges.
Keywords:
First Keyword, Second Keyword, Third Keyword
Proceedings Editor
Edmond Hajrizi
ISBN
978-9951-982-41-2
Location
UBT Lipjan, Kosovo
Start Date
25-10-2025 9:00 AM
End Date
26-10-202 6:00 PM
DOI
10.33107/ubt-ic.2025.93
Recommended Citation
Xheladini, Blerta; Sofiu, Vehebi; Qehaja, Besnik; Kabashi, Faton; Shkurti, Lamir; and Selimaj, Mirlinda, "Analysis of Cyber Risk Awareness Through Classification Algorithms" (2025). UBT International Conference. 25.
https://knowledgecenter.ubt-uni.net/conference/2025UBTIC/CS/25
Analysis of Cyber Risk Awareness Through Classification Algorithms
UBT Lipjan, Kosovo
Exposure to cyber risks has increased significantly with the development of technology. Sophisticated attacks such as phishing, ransomware, malware, and credential misuse represent serious threats, while the limited awareness of technology users makes them highly vulnerable. This thesis develops along two parallel dimensions: assessing the level of cyber risk awareness among the citizens of Prizren and analyzing the potential offered by classification algorithms in preventing cyber-attacks. Data on citizens’ knowledge, practices, and perceptions of cybersecurity were collected through a questionnaire. The analysis of results revealed an acceptable level of awareness regarding the general concept of cyber risk; however, knowledge about specific threats was found to be limited. The thesis is divided into six chapters, including a literature review and methodology. In the technical part, the study examines the use of classification algorithms in ma-chine learning – including logistic regression, decision trees, random forest, and neural networks – with a special focus on XGBoost. This algorithm, known for its accuracy and flexibility, was used to automatically classify open-ended questionnaire responses, demonstrating its ability to categorize textual data into categories such as suggestions, issues, or lack of response. The results showed high accuracy and efficiency, making XGBoost a very reliable tool for practical and educational analyses. The final findings highlight the importance of integrating advanced classification algorithms into educational and protective systems. This thesis contributes to the scientific literature by combining empirical analysis with modern machine learning methods, providing a foundation for developing more effective strategies to tackle contemporary cyber challenges.