Skip to main content

Table 6 Another Example with Real Data

From: Thinking outside the curve, part I: modeling birthweight distribution

Model

Number of

FLIC votes

Fitted Mixture Density

2 components

0

.144 f(x;2533,1031) + .856 f(x;3241,452)

3 components

0

.040 f(x;1300,487) + .833 f(x;3215,450) + .127 f(x;3427,656)

4 components

1

.012 f(x;778,186) + .100 f(x;2292,683) + .760 f(x;3198,419) +

.128 f(x;3668,511)

5 components

1

.010 f(x;730,153) + .043 f(x;1700,490) + .655 f(x;3200,435) +

.282 f(x;3289,538) + .010 f(x;4175,439)

6 components

22

.007 f(x;651,103) + .015 f(x;1137,273) + .213 f(x;2815,666) +

.638 f(x;3191,379) + .116 f(x;3747,361) + .011 f(x;4340,409)

7 components

1

.007 f(x;645,101) + .013 f(x;1083,246) + .108 f(x;2415,574) +

.496 f(x;3091,372) + .332 f(x;3456,383) + .038 f(x;4021,341)

+.006 f(x;4613,347)

  1. Parameters in 2-component through 7-component normal mixture models for birthweight distribution are estimated, based on 25 samples of size 50000 from the population of black singletons in general. The numbers of samples for which the FLIC preferred the various models are also recorded.