Bayesian models have been popular in the domains of estimation and forecasting in several areas. Extensions to such models include the integration of support vector machines in efforts to introduce a higher level of specificity to the model, and develop what is known by the Bayesian kernel models. These new models are developed for the Gaussian distribution. In this paper, we extend the model to analyze count data and develop a Bayesian kernel model in which we use the Gamma conjugate prior with Poisson likelihood function. This model is deployed in an analysis of the frequency of disruptive events, especially applied to disruptions of the links comprising the inland waterway transportation network of the Mississippi River Navigation System. Ultimately such analysis can forecast the occurrence of disruptive events and improve preparedness and recovery decision making under uncertainty.