Cascading blackouts in power systems have an immense impact on the society in many ways. Therefore, the modeling and analysis of cascading failures in a large-scale power grid interconnection have attracted high interest from the power systems research community. In this report, we present a modeling framework to predict the propagation sequence of cascades outages for the given initial failures. Specifically, a graph-based representation is introduced to transform the impact of existing failures in the system to the rest of system components, while approaching the probability of failure events from the perspective of survival theory. To construct the impact graph, we have analyzed the distribution of the size and duration in the generated dataset of cascading outage events and determined the time shape functions that best fit the dataset. This analysis has led to obtaining the hidden graph over which outages propagate. The resultant propagation graph improves the understanding of how outages influence other systems components, allows for predicting the cascading outage sequence, as well as potentially facilities the design of mitigation strategies for tackling the cascading outage problems.