Development of Crash Prediction Models for Transportation Planning Analysis.

摘要

Transportation planning is a vital and necessary operation for a metropolitan area to grow. As such, and in order to receive Federal funding for transportation projects, metropolitan areas engage in transportation planning as regulated by MAP-21. One element of meeting MAP-21 requirements is addressing the safety of a region. With new requirements by MAP-21, MPOs must demonstrate some sort of performance measure showing changes in the various elements, making quantitative means of displaying these changes ever more important.;The goal of this project was to develop a model or set of models that could produce quantitative results as opposed to the traditional qualitative results gained from professional opinion. This allows for better decision-making for test project scoring in transportation plans and additionally in meeting MAP-21 requirements. Following a review of the current available methodologies and an inventory of other states' efforts to develop crash prediction models, mathematical modeling for Louisiana statewide crash prediction formulae were attempted. These results and the methodology were deemed unsuitable for the desired outcomes and the use of SVR modeling was explored.;The use of the SVR models described in this report produce acceptable results, have been validated for use in forecasting, and allow for the comparison of conditions between base data, future years, and future years with MTP test projects included. The results of these models provide transportation planners increased means to determine project rankings based on safety as well as satisfy MAP-21 requirements.