| Class | Analysis and data manipulation command |
| Name | mixture |
| Arguments | <quantitative trait> [<Number of distributions> [normal|pooled_normal|exponential|poisson]] |
Estimate mixing proportions, means and standard deviations for a 1..5 component mixture model describing the specified quantititative trait. The default is a mixture of Normal (Gaussian) distributions with different means and variances, but a common variance can alternatively be specified. Other distributions available are the exponential and Poisson. A line-printer type histogram is produced.
Example:
# A mixture of two normals from:
# Everitt B.S. and Hand D.J. (1981) Finite mixture distributions.
# Chapman and Hall. p.46.
#
# means: 19.96 26.16
# variances: 4.6225 7.6176
# proportions: 0.65 0.35
#
>> macro mixexample
>> if (index > %runtot and index <= (%runtot + %3)) then %1=%2
>> eval (define runtot (+ runtot %3))
>> ;;;;
>>
>> set loc xvar
>> sim ped 1000 1 1
>> run
>> eval (define runtot 0)
>> mixexample xvar 15.5 10
>> mixexample xvar 16.5 21
>> mixexample xvar 17.5 56
>> mixexample xvar 18.5 79
>> mixexample xvar 19.5 114
>> mixexample xvar 20.5 122
>> mixexample xvar 21.5 110
>> mixexample xvar 22.5 85
>> mixexample xvar 23.5 85
>> mixexample xvar 24.5 61
>> mixexample xvar 25.5 47
>> mixexample xvar 26.5 49
>> mixexample xvar 27.5 47
>> mixexample xvar 28.5 44
>> mixexample xvar 29.5 31
>> mixexample xvar 30.5 20
>> mixexample xvar 31.5 11
>> mixexample xvar 32.5 4
>> mixexample xvar 33.5 4
>>
>> mix xvar 1
>> mix xvar 2
------------------------------------------------
Mixture distributions for trait "xval"
------------------------------------------------
Intvl Midpt Count Histogram
-------------------------------------------------------
15.5000 10 *
16.5000 21 **
17.5000 56 *****
18.5000 79 *******
19.5000 114 | ***********
20.5000 122 | ************
21.5000 110 + ***********
22.5000 85 | ********
23.5000 85 | ********
24.5000 61 | ******
25.5000 47 | ****
26.5000 49 ****
27.5000 47 ****
28.5000 44 ****
29.5000 31 ***
30.5000 20 **
31.5000 11 *
32.5000 4
33.0263 0
33.5000 4
Filliben correlation = 0.2732 (P=0.000)
Poissonness test Z = -2477.2956 (P= NaN)
Median (IQR) = 21.5000 ( 19.5000 -- 25.5000)
Symmetry test J(.02) = 0.3333 (P=0.000)
Distribution type = Normal
No. of distributions = 2
No. of observations = 1000
No. of unique values = 19
-2*Loglikelihood = 3536.9155
Dist Mean Standard Dev Proportion
---------------------------------------------
1 20.4548 2.1459 0.6518
2 26.6588 2.7603 0.3482
>> lrt
Term -2*LL NPar P-value
------ ----------- ---- -------
Model0 3666.5179 2
Model1 3536.9155 4
LRTS 129.6024 2 0.0000
| << (clreg) | Up to index | >> (lifetable) |