Simple Methodology for Predicting Cyber-Attacks

by Rong Chen, Giuseppe Corvasce, Ruey S. Tsay

The paper proposes a simple statistical methodology for predicting and forecasting cyber-attacks. The procedure models the duration, defined as the number of days from one cyber-attack to the next ones, relying on the ARMA(1,1)/PARCH(1,1) specification, with the aim to also describe the evolution of the conditional volatility of the duration. The empirical analysis is based on disruptive, exploitive as well as mixed and undetermined cyber-attacks perpetrated by criminals and reports in sample and out-of-sample results of the statistical methodology, showing several metrics of performance and goodness of fit.

Contact Prof. Dr. Giuseppe Corvasce with questions about implementation.