A Clustering Approach to Identify High-Risk Taxi Drivers Based on Self-Reported Driving Behavior

Table 7

First Logistic regression model.

Model	Step	Predictor	B	-value	Odds	-2LL	CHI2
High risk drivers vs. low/medium risk drivers	1	Annual mileage	1.69	0.000	1.18	444.27	28.62
	2	Lapses	0.48	0.109	1.61	276.16	196.72
		Inattention errors	0.83	0.200	2.32
		Ordinary violations	1.32	0.000	3.76
		Inexperience errors	−0.31	0.190	0.66
		Aggressive violations	3.34	0.000	28.26