Application of naïve Bayesian approach in detecting reproducible fatal collision locations on freeway.

摘要: Detecting high-collision-concentration locations based solely on collision frequency may produce different results compared to those considering the severities of collisions. In particular, it can lead government agencies focusing sites with a high while neglecting lower but higher percentage injury and fatal This study developed systematic ways detecting reproducible (R) using naive Bayes approach continuous risk profile (CRP) that estimates true by filtering out random noise in data. The posterior probability collisions being at location is estimated relationship between spatial distribution fatal-collision (i.e., likelihood) CRP prior probability). proposed method be used detect highest proxy measure (PMP) observing R. An empirical evaluation 5-year traffic data from six routes California shows R PMP outperform SPF-based approaches or selection, regardless various conditions parameters method. only requires annual volume estimate prioritize PMPs across multiple routes. Therefore, helps prioritizing where number reduced, thus help them save lives limited resources collection.