Stacked survival models for residual lifetime data

Table 3 Mean weights for a stacked model including all submodels or including all submodels except Weibull. The failure time data were generated according to a Weibull (2, 2) distribution with varying amounts of administrative censoring for samples of size 125 over 100 simulation runs

	Individual submodel type of stacked estimator
Administrative Censoring Proportion	NPMLE	Weibull	Log-Logistic	Log-Normal	Gamma
10%	0.01394	0.08208	0.02781	0.04167	0.09579
	0.01607	N/A	0.05645	0.08028	0.8472
20%	5.586×10⁻⁹	0.9061	0.01439	0.04962	0.02992
	8.419×10⁻⁹	N/A	0.04088	0.08901	0.8701
30%	3.900×10⁻⁹	0.8503	0.03581	0.04999	0.06389
	5.520×10⁻⁹	N/A	0.07797	0.08018	0.8418
40%	2.169×10⁻⁹	0.7888	0.02656	0.01465	0.1700
	3.463×10⁻⁹	N/A	0.1319	0.04835	0.8197

ISSN: 1471-2288