Adding (2) kinds of error checking to RAM and ML objectives with covariance data...
[openmx:openmx.git] / demo / OneFactorModel_PathCov.R
1 #
2 #   Copyright 2007-2010 The OpenMx Project
3 #
4 #   Licensed under the Apache License, Version 2.0 (the "License");
5 #   you may not use this file except in compliance with the License.
6 #   You may obtain a copy of the License at
7
8 #        http://www.apache.org/licenses/LICENSE-2.0
9
10 #   Unless required by applicable law or agreed to in writing, software
11 #   distributed under the License is distributed on an "AS IS" BASIS,
12 #   WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
13 #   See the License for the specific language governing permissions and
14 #   limitations under the License.
15
16
17 # -----------------------------------------------------------------------
18 # Program: OneFactorModel_PathCov.R  
19 #  Author: Ryne Estabrook
20 #    Date: 08 01 2009 
21 #
22 # One Factor model to estimate factor loadings, residual variances and means
23 # Path style model input - Covariance matrix data input
24 #
25 # Revision History
26 #   Hermine Maes -- 10 08 2009 updated & reformatted
27 # -----------------------------------------------------------------------
28
29 require(OpenMx)
30
31 #Prepare Data
32 # -----------------------------------------------------------------------
33 myFADataCov<-matrix(
34         c(0.997, 0.642, 0.611, 0.672, 0.637, 0.677,
35           0.642, 1.025, 0.608, 0.668, 0.643, 0.676,
36           0.611, 0.608, 0.984, 0.633, 0.657, 0.626,
37           0.672, 0.668, 0.633, 1.003, 0.676, 0.665,
38           0.637, 0.643, 0.657, 0.676, 1.028, 0.654,
39           0.677, 0.676, 0.626, 0.665, 0.654, 1.020),
40         nrow=6,
41         dimnames=list(
42                 c("x1","x2","x3","x4","x5","x6"),
43                 c("x1","x2","x3","x4","x5","x6"))
44 )
45
46 myFADataMeans <- c(2.988, 3.011, 2.986, 3.053, 3.016, 3.010)
47 names(myFADataMeans) <- c("x1","x2","x3","x4","x5","x6")
48
49 #Create an MxModel object
50 # -----------------------------------------------------------------------
51 oneFactorModel <- mxModel("Common Factor Model Path Specification", 
52         type="RAM",
53         mxData(
54                 observed=myFADataCov, 
55                 type="cov", 
56                 numObs=500,
57                 mean=myFADataMeans
58         ),
59         manifestVars=c("x1","x2","x3","x4","x5","x6"),
60         latentVars="F1",
61         # residual variances
62         mxPath(
63                 from=c("x1","x2","x3","x4","x5","x6"),
64                 arrows=2,
65                 free=TRUE,
66                 values=c(1,1,1,1,1,1),
67                 labels=c("e1","e2","e3","e4","e5","e6")
68         ),
69         # latent variance
70         mxPath(from="F1",
71                 arrows=2,
72                 free=TRUE,
73                 values=1,
74                 labels ="varF1"
75         ),
76         # factor loadings
77         mxPath(from="F1",
78                 to=c("x1","x2","x3","x4","x5","x6"),
79                 arrows=1,
80                 free=c(FALSE,TRUE,TRUE,TRUE,TRUE,TRUE),
81                 values=c(1,1,1,1,1,1),
82                 labels =c("l1","l2","l3","l4","l5","l6")
83         ), 
84         # means
85         mxPath(from="one",
86                 to=c("x1","x2","x3","x4","x5","x6","F1"),
87                 arrows=1,
88                 free=c(TRUE,TRUE,TRUE,TRUE,TRUE,TRUE,FALSE),
89                 values=c(1,1,1,1,1,1,0),
90                 labels =c("meanx1","meanx2","meanx3","meanx4","meanx5","meanx6",NA)
91         ) 
92 ) # close model
93
94 oneFactorFit <- mxRun(oneFactorModel)
95
96 summary(oneFactorFit)
97 oneFactorFit@output$estimate
98
99 #Compare OpenMx results to Mx results 
100 # -----------------------------------------------------------------------
101 omxCheckCloseEnough(oneFactorFit@output$estimate[["l2"]], 0.999, 0.01)
102 omxCheckCloseEnough(oneFactorFit@output$estimate[["l3"]], 0.959, 0.01)
103 omxCheckCloseEnough(oneFactorFit@output$estimate[["l4"]], 1.028, 0.01)
104 omxCheckCloseEnough(oneFactorFit@output$estimate[["l5"]], 1.008, 0.01)
105 omxCheckCloseEnough(oneFactorFit@output$estimate[["l6"]], 1.021, 0.01)
106 omxCheckCloseEnough(oneFactorFit@output$estimate[["varF1"]], 0.645, 0.01)
107 omxCheckCloseEnough(oneFactorFit@output$estimate[["e1"]], 0.350, 0.01)
108 omxCheckCloseEnough(oneFactorFit@output$estimate[["e2"]], 0.379, 0.01)
109 omxCheckCloseEnough(oneFactorFit@output$estimate[["e3"]], 0.389, 0.01)
110 omxCheckCloseEnough(oneFactorFit@output$estimate[["e4"]], 0.320, 0.01)
111 omxCheckCloseEnough(oneFactorFit@output$estimate[["e5"]], 0.370, 0.01)
112 omxCheckCloseEnough(oneFactorFit@output$estimate[["e6"]], 0.346, 0.01)
113 omxCheckCloseEnough(oneFactorFit@output$estimate[["meanx1"]], 2.988, 0.01)
114 omxCheckCloseEnough(oneFactorFit@output$estimate[["meanx2"]], 3.011, 0.01)
115 omxCheckCloseEnough(oneFactorFit@output$estimate[["meanx3"]], 2.986, 0.01)
116 omxCheckCloseEnough(oneFactorFit@output$estimate[["meanx4"]], 3.053, 0.01)
117 omxCheckCloseEnough(oneFactorFit@output$estimate[["meanx5"]], 3.016, 0.01)
118 omxCheckCloseEnough(oneFactorFit@output$estimate[["meanx6"]], 3.010, 0.01)