Keywords - Function groups - @ A B C D E F G H I J K L M N O P Q R S T U V W X Y Z

Library: stats
See also: CPC draftman factor pca CPCprop CPCp CPCFGalg Jacobirot

Quantlet: FluryStepUp
Description: computes sequentially all possible CPC and CPCp models: Beginning with the proportional model, it steps down to the full CPC model and estimates subsequently all possible CPCp models. Additionally to Chi-Square-Statistics, it provides the Akaike (AIC) and Schwarz (SIC) Information Criteria for model selection.

Reference(s):

Usage: FluryStepUp(A,N)
Input:
A p x p x k array, consisting of k covariance matrices (p x p)
N k x 1 vector of weights, usually number of observations in group k

Note:

Example:
library("stats")
;Input of grouped covariances
covar1=#(112.01, 106.64, 52.97)~#(106.64,108.13,54.75)~#(52.97,54.75,33.86)
covar2=#(86.08,81.66,40.24)~#(81.66,85.54,42.08)~#(40.24,42.08,26.66)
covar3=#(65.4,60.23,24.69)~#(60.23,62.27,23.47)~#(24.69,23.47,16.33)
covar4=#(88.66,79.11,41.32)~#(79.11,80.57,38.81)~#(41.32,38.81,23.97)
covar=reshape(covar1~covar2~covar3~covar4,#(3,3,4))
;Sample size
n=#(173,141,88,76)
FluryStepUp(covar,n)

Result:
Contents of _tmp
[1,] "proportional Model done"

Contents of _tmp
[1,] "CPC Model done"

Contents of _tmp
[1,] "CPCp Model done q = 1"

Contents of Output
[ 1,] "The Flury StepUp Approach for Model Selection in CPC - Models                           "
[ 2,] "----------------------------------------------------------------------------------------"
[ 3,] "----------------------------------------------------------------------------------------"
[ 4,] "------Model-----------------------------------------------------------------------------"
[ 5,] "Higher                   Lower             Chi2         df      p-val      AIC   Schwarz"
[ 6,] "----------------------------------------------------------------------------------------"
[ 7,] "Equality            Proportionality        0.772         3      0.86      53.9      53.9"
[ 8,] "Proportionality          CPC              37.626         6      0.00      59.1      71.6"
[ 9,] "CPC                   CPCp( 1)             8.497         3      0.04      33.5      71.0"
[10,] "CPCp( 1)             Unrelated             6.974         6      0.32      31.0      81.0"
[11,] "Unrelated                ---                                              36.0     111.1"



Author: M. R. Fengler, 20010725 license MD*Tech
(C) MD*TECH Method and Data Technologies, 05.02.2006