Variability of boat traffic has been a central concern of the model formulation. As the model was being formed, there was no working definition of what ``variable'' meant. The reason for the data-mining techniques described in appendix A was to find relationships between boaters and destinations. The low explanatory power of the relationships attests to the difficulty in establishing correlations. Further work on these datasets, and future data collection efforts specifically designed with more direct hypotheses in mind would be valuable in improving the inferences that can be drawn from such a dataset.
While trip length was not used in the simulation as is presented in this study, it can be a useful proxy variable for the range of destinations that a boater would select at embarkation. If a boat in Coral Shores (at the northernmost part of the bay) has an average trip length of three hours, that would give it a very low probability of selecting Venice or Punta Gorda as a destination, which could mean a trip of several hours. This information was not used in the calibration phase of this simulation, though it's use in future studies may yield useful inferences about how boat types and destinations match up to each other. Independence was assumed between trip length and the probability that the boat would embark: that assumption was justified by the available data.