Table 10:

Gamma simulation, censored (full results). For categorical model with interpolation, the d-cal is already very low at λ = 0 so it is hard to optimize this one further.

	λ	0	1	5	10	50	100	500	1000
Log-Norm nll	nll	−0.059	−0.049	−0.022	0.004	0.099	0.138	0.191	0.215
	d-cal	0.029	0.020	0.008	0.005	7e-4	2e-4	6e-5	7e-5
	conc	0.981	0.969	0.950	0.942	0.927	0.916	0.914	0.897
Log-Norm s-crps	nll	0.038	0.084	0.119	0.143	0.185	0.201	0.343	0.436
	d-cal	0.017	0.007	0.003	0.001	1e-4	1e-4	5e-5	8e-5
	conc	0.982	0.978	0.971	0.963	0.952	0.950	0.850	0.855
Cat-ni	nll	0.797	0.799	0.805	0.822	1.023	1.149	1.665	1.920
	d-cal	0.009	0.006	0.003	0.002	3e-4	2e-4	6e-5	6e-5
	conc	0.987	0.987	0.987	0.987	0.982	0.976	0.922	0.861
Cat-i	nll	0.783	0.782	0.788	0.795	0.948	1.124	1.686	1.994
	d-cal	7e-5	1e-4	6e-5	8e-5	2e-4	2e-4	4e-5	6e-5
	conc	0.987	0.987	0.987	0.987	0.983	0.976	0.933	0.847
mtlr-NI	nll	0.873	0.875	0.875	0.977	1.271	1.412	1.747	1.900
	d-cal	0.004	0.004	0.003	0.004	4e-4	2e-4	2e-4	2e-4
	conc	0.987	0.987	0.987	0.985	0.973	0.965	0.951	0.943
mtlr-I	nll	0.829	0.830	0.866	0.981	1.266	1.414	1.762	1.912
	d-cal	0.004	0.004	0.004	0.004	5e-4	1e-4	6e-5	7e-5
	conc	0.988	0.988	0.987	0.985	0.971	0.963	0.947	0.939