An Automated Profile-Likelihood-Based Algorithm for Fast Computation of the Maximum Likelihood Estimate in a Statistical Model for Crash Data

Table 2

Results for Scenario 1 (values in brackets are standard deviations).

		PLBA	MM	NR
		0.864 (0.180)	0.864 (0.180)	0.878 (0.181)
		0.515 (0.071)	0.515 (0.071)	0.513 (0.071)
		0.310 (0.066)	0.310 (0.066)	0.309 (0.067)
		0.175 (0.054)	0.175 (0.054)	0.179 (0.054)
		0.248 (0.062)	0.248 (0.062)	0.249 (0.063)
		0.453 (0.068)	0.453 (0.068)	0.448 (0.068)
		0.299 (0.064)	0.299 (0.064)	0.302 (0.064)
	Convergence proportion (%)	100	100	54.3
	Iterations	5.7 (0.5)	27.8 (5.6)	7 (1.2)
	CPU time (secs)	0.0005	0.0043	0.0029
	Time ratio	1	8	6
	Log-likelihood	191.95	191.95	191.95
	MSE
		0.851 (0.017)	0.851 (0.017)	0.852 (0.017)
		0.520 (0.007)	0.520 (0.007)	0.520 (0.007)
		0.310 (0.007)	0.310 (0.007)	0.310 (0.007)
		0.170 (0.005)	0.170 (0.005)	0.170 (0.005)
		0.250 (0.006)	0.250 (0.006)	0.250 (0.006)
		0.450 (0.007)	0.450 (0.007)	0.450 (0.007)
		0.300 (0.007)	0.300 (0.007)	0.300 (0.007)
	Convergence proportion (%)	100	100	55.9
	Iterations	6 (0.1)	36.3 (4.5)	7 (1.3)
	CPU time (secs)	0.0004	0.0056	0.0030
	Time ratio	1	14	7
	Log-likelihood	19380	19380	19380
	MSE