A Fast and Efficient Estimation of the Parameters of a Model of Accident Frequencies via an MM Algorithm

Table 4

Results for Scenario 4 ( and ). Values in brackets are standard deviations.

		MM	SqS3	NR	BFGS
	Convergence proportion (%)	100	100	98.9	21.1
	Iterations	32.8 (3.9)	8.3 (0.8)	11.4 (4)	13.8 (0.5)
	CPU time (secs)	0.010	0.004	0.089	0.454
	Time ratio	1.00	0.46	9.33	47.45
	Log-likelihood	-1046.83	-1046.83	-1046.88	-1047.3
	MSE
	Convergence proportion (%)	100	100	99.4	24.8
	Iterations	43.9 (4.1)	9.9 (0.9)	11.5 (3.9)	15.8 (5.5)
	CPU time (secs)	0.014	0.006	0.090	0.576
	Time ratio	1.00	0.41	6.32	40.50
	Log-likelihood	-104802.55	-104802.55	-104800.93	-108716.18
	MSE