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

Table 5

Results for Scenario 5 ( and ). Values in brackets are standard deviations.

		MM	SqS3	NR	BFGS
	Convergence proportion (%)	100	100	97.9	4.1
	Iterations	34.9 (3.5)	8.4 (0.7)	14.3 (4.9)	14 (0)
	CPU time (secs)	0.017	0.007	0.338	1.162
	Time ratio	1.00	0.42	19.54	67.25
	Log-likelihood	-2045.46	-2045.46	-2045.65	-2046.96
	MSE
	Convergence proportion (%)	100	100	96.8	16.2
	Iterations	46.3 (3.6)	10.1 (0.8)	14.9 (5.2)	6.6 (7.2)
	CPU time (secs)	0.018	0.006	0.347	0.582
	Time ratio	1.00	0.35	19.25	32.31
	Log-likelihood	-203353.48	-203353.48	-203355.72	-245597.55
	MSE