A Machine Learning Approach For Predicting Vulnerability Assessment Requests

Within the last years, the amount of data collected has been growing exponentially. It is important to start to take advantage of this growth by using this huge amount of data in order to develop new solutions that can improve our daily life. In the same way, time series prediction has been attracting the interest of a lot of researchers due to its increasing importance in decision making activities in many fields of knowledge. The motivation behind this thesis comes from the activities held during my internship in a financial institution. During the months of my internship, I was allocated in the department of Cyber Security Protection. The main task that I was responsible of was handling the requests of the security assessments. In a few words, I was handling the new security requests coming from the Application Owners of the institution and then I would carry the vulnerability scans and the needed penetration tests in order to make sure that the application and all of its assets were secured and no vulnerabilities were present. Having that, this financial institution has global measures, I was receiving quite a few security assessment requests daily. Therefore, I noticed the need for a better organization of the work load. The aim of my thesis is to predict the upcoming security assessment requests in order to have a better organization of the work load. Also, this approach, will not only help me to organize better my work load but also to prioritize those requests that are more critical. In this thesis, I have implemented the most used predictive models mentioned in the research literature, in order to test these models in a real-life problem.