The FMM Procedure

Comparison with Roeder’s Method

It is important to note that Roeder’s original analysis proceeds in a different manner than the finite mixture modeling presented here. The technique presented by Roeder first develops a "best" range of scale parameters based on a specific criterion. Roeder then uses fixed scale parameters taken from this range to develop optimal equal-scale Gaussian mixture models.

You can reproduce Roeder’s point estimate for the density by specifying a five-component Gaussian mixture. In addition, use the EQUATE=SCALE option in the MODEL statement and a RESTRICT statement fixing the first component’s scale parameter at 0.9025 (Roeder’s h = 0.95, scale$=h^2$). The combination of these options produces a mixture of five Gaussian components, each with variance 0.9025. The following statements conduct this analysis:

title2 "Five Components, Equal Variances = 0.9025";
proc fmm data=galaxies;
   model v = / K=5 equate=scale;
   restrict int 0 (scale 1) = 0.9025;
run;

The output is shown in Figure 39.19 and Figure 39.20.

Figure 39.19: Reproduction of Roeder’s Five-Component Analysis of Galaxy Data

FMM Analysis of Galaxies Data
Five Components, Equal Variances = 0.9025

The FMM Procedure

Model Information
Data Set WORK.GALAXIES
Response Variable v
Type of Model Homogeneous Mixture
Distribution Normal
Components 5
Link Function Identity
Estimation Method Maximum Likelihood

Fit Statistics
-2 Log Likelihood 412.2
AIC (Smaller is Better) 430.2
AICC (Smaller is Better) 432.7
BIC (Smaller is Better) 451.9
Pearson Statistic 82.5549
Effective Parameters 9
Effective Components 5

Linear Constraints at Solution
k = 1   Constraint
Active
Variance = 0.90 Yes

Parameter Estimates for Normal Model
Component Parameter Estimate Standard
Error
z Value Pr > |z|
1 Intercept 26.3266 0.7778 33.85 <.0001
2 Intercept 33.0443 0.5485 60.25 <.0001
3 Intercept 9.7101 0.3591 27.04 <.0001
4 Intercept 23.0295 0.2294 100.38 <.0001
5 Intercept 19.7187 0.1784 110.55 <.0001
1 Variance 0.9025 0    
2 Variance 0.9025 0    
3 Variance 0.9025 0    
4 Variance 0.9025 0    
5 Variance 0.9025 0    

Parameter Estimates for Mixing Probabilities
Component
Mixing
Probability
Linked Scale
GLogit(Prob) Standard
Error
z Value Pr > |z|
1 0.0397 -2.4739 0.7084 -3.49 0.0005
2 0.0366 -2.5544 0.6016 -4.25 <.0001
3 0.0854 -1.7071 0.4141 -4.12 <.0001
4 0.3678 -0.2466 0.2699 -0.91 0.3609
5 0.4706 0      



Figure 39.20: Density Plot for Roeder’s Analysis

 Density Plot for Roeder’s Analysis