Sensitivity of a real-time freeway crash prediction model to calibration optimality

European Transport Research Review

An Open Access Journal

Table 1 Comparing ad hoc and near-optimal methods

Method	Crash Precursor	Categorization^a	Boundary Values (B1, B2, B3)
Ad Hoc	CVS	20/30/30/20	(0.062,0.089,0.139)
	Q	20/30/30/20	(−9.19,0.09,8.77)
	COVV	40/40/20	(1.49,3.44,−)
Near-Optimal	CVS	37/33/30	(0.079,0.126,−)
	Q	43/37/20	(−1.433,10.05,−)
	COVV	23/34/43	(0.815,1.833,−)

^aNumbers in this column represent the percentage share of each category. For instance 40/40/20 means 40% of the observations for the precursor fall into the first category, 40% fall into in the second category and 20% fall into the third category