Time series modeling of vulnerabilities

Document Type


Publication Date



Computer Information Systems


Vulnerability prediction models forecast future vulnerabilities and can be used to assess security risks and estimate the resources needed for handling potential security breaches. Although several vulnerability prediction models have been proposed, such models have shortcomings and do not consider trend, level, and seasonality components of vulnerabilities. Through time series analysis, this study built predictive models for five popular web browsers: Chrome, Firefox, Internet Explorer, Safari and Opera and for all reported vulnerabilities elsewhere. Results showed that time series models provide a good fit to our vulnerability datasets and can be useful for vulnerability prediction. Results also suggested that the level of the series is the best estimator of the prediction models.