ABSTRACT
Theory and application of rare events have been very important in recent years due to its practical importance in very different fields such as insurance, finance, engineering or environmental science. This paper presents a methodology for predicting rare events based on Bayesian Networks which in turn enables the study alternative scenarios to control the frequency of road accidents. This way the model Naive-Poisson and ROCDM is presented in this paper for its validation. The developed model is used to estimate and predict road accidents as rare events and results have been evaluated by using ROCDM curve. Naive-Poisson model and a validation model based on ROC curve is used to study several Spanish roads and the results are here shown.
Support the magazine and subscribe to the content
This is premium stuff. Subscribe to read the entire article.