[Depmix-commits] r38 - trunk
noreply at r-forge.r-project.org
noreply at r-forge.r-project.org
Tue Mar 4 23:20:30 CET 2008
Author: ingmarvisser
Date: 2008-03-04 23:20:30 +0100 (Tue, 04 Mar 2008)
New Revision: 38
Modified:
trunk/depmix-test1balance2.R
Log:
Added working example of contraints on balance scale data model and usage of likelihood ratio test
Modified: trunk/depmix-test1balance2.R
===================================================================
--- trunk/depmix-test1balance2.R 2008-03-04 22:16:47 UTC (rev 37)
+++ trunk/depmix-test1balance2.R 2008-03-04 22:20:30 UTC (rev 38)
@@ -2,12 +2,9 @@
#
# Started by Ingmar Visser 26-2-2008
#
-# Usage: go to trunk directory and source this file in R, if the program
-# still works it should return TRUE at every test (or make immediate sense
-# otherwise)
+# Usage: go to trunk directory and source this file in R
+#
-# Changes:
-
#
# BALANCE SCALE data example with age as covariate on class membership
#
@@ -64,6 +61,7 @@
llratio(mod2,mod1)
+
predict(mod2 at response[[1]][[1]])[1,]
predict(mod2 at response[[1]][[2]])[1,]
predict(mod2 at response[[1]][[3]])[1,]
@@ -99,447 +97,3 @@
-# trichotome data
-dat3 <- dat[,c(4:7)]
-dat3 <- markovdata(dat3,nt=rep(1,nrow(dat3)),itemt=c(3,3,3,3))
-
-
-#
-# TRICHOTOME DATA MODELLEN
-#
-
-lc1 <- lca(nc=1,itemt=c(3,3,3,3))
-fit1 <- fitdmm(dat3,lc1)
-
-lc2 <- lca(nc=2,itemt=c(3,3,3,3))
-fit2 <- fitdmm(dat3,lc2)
-
-lc3 <- lca(nc=3,itemt=c(3,3,3,3))
-fit3.2 <- fitdmm(dat3,lc3)
-
-lc4 <- lca(nc=4,itemt=c(3,3,3,3))
-fit4 <- fitdmm(dat3,lc4)
-
-
-
-Model: 1 class model fitted at Mon Feb 18 11:48:00 2008
-Optimization information, method is donlp
- Iterations: 37
- Inform: KT-conditions satisfied, no further correction computed (look up the respective manuals for more information.)
-
- Loglikelihood of fitted model: -2227.966
- AIC: 4471.932
- BIC: 4509.196
- Number of observations (used in BIC): 779
- Fitted model
- Model: 1 class model
- Number of parameters: 15
- Free parameters: 8
- Number of states: 1
- Number of items: 4
- Item types: 3 3 3 3
-
- Parameter values, observation parameters
-
- Item1,p 1 Item1,p 2 Item1,p 3 Item2,p 1 Item2,p 2 Item2,p 3 Item3,p 1 Item3,p 2
-Class1 0.659 0.322 0.019 0.018 0.302 0.680 0.653 0.309
-se 0.017 0.017 0.005 0.005 0.016 0.017 0.017 0.017
-t 38.760 19.244 3.911 3.776 18.344 40.720 38.322 18.680
- Item3,p 3 Item4,p 1 Item4,p 2 Item4,p 3
-Class1 0.037 0.017 0.294 0.689
-se 0.007 0.005 0.016 0.017
-t 5.488 3.636 18.010 41.577
-
-
- Parameter values, unconditional (class) probabilities
-
- Class1
-val 1
-se 0
-t NA
-
-
-Model: 2 class model fitted at Mon Feb 18 11:50:22 2008
-Optimization information, method is donlp
- Iterations: 30
- Inform: KT-conditions satisfied, no further correction computed (look up the respective manuals for more information.)
-
- Loglikelihood of fitted model: -1326.523
- AIC: 2687.046
- BIC: 2766.232
- Number of observations (used in BIC): 779
- Fitted model
- Model: 2 class model
- Number of parameters: 31
- Free parameters: 17
- Number of states: 2
- Number of items: 4
- Item types: 3 3 3 3
-
- Parameter values, observation parameters
-
- Item1,p 1 Item1,p 2 Item1,p 3 Item2,p 1 Item2,p 2 Item2,p 3 Item3,p 1 Item3,p 2
-Class1 0.036 0.919 0.045 0.037 0.892 0.070 0.030 0.895
-se 0.015 0.020 0.014 0.012 0.023 0.021 0.013 0.021
-t 2.393 46.329 3.293 3.021 38.103 3.392 2.270 42.384
-Class2 0.938 0.055 0.008 0.009 0.037 0.954 0.933 0.047
-se 0.012 0.011 0.004 0.004 0.009 0.010 0.013 0.011
-t 78.442 4.879 1.875 2.185 4.086 95.947 73.884 4.270
- Item3,p 3 Item4,p 1 Item4,p 2 Item4,p 3
-Class1 0.075 0.041 0.893 0.067
-se 0.017 0.013 0.024 0.022
-t 4.354 3.151 36.567 3.083
-Class2 0.020 0.006 0.026 0.968
-se 0.006 0.003 0.007 0.008
-t 3.215 1.718 3.423 117.896
-
-
- Parameter values, unconditional (class) probabilities
-
- Class1 Class2
-val 0.309 0.691
-se 0.018 0.018
-t 17.535 39.123
-
-Model: 3 class model fitted at Mon Feb 18 11:54:10 2008
-Optimization information, method is donlp
- Iterations: 104
- Inform: KT-conditions satisfied, no further correction computed (look up the respective manuals for more information.)
-
- Loglikelihood of fitted model: -1243.951
- AIC: 2539.902
- BIC: 2661.01
- Number of observations (used in BIC): 779
- Fitted model
- Model: 3 class model
- Number of parameters: 49
- Free parameters: 26
- Number of states: 3
- Number of items: 4
- Item types: 3 3 3 3
-
- Parameter values, observation parameters
-
- Item1,p 1 Item1,p 2 Item1,p 3 Item2,p 1 Item2,p 2 Item2,p 3 Item3,p 1 Item3,p 2
-Class1 0.938 0.056 0.006 0.009 0.038 0.953 0.935 0.048
-se 0.012 0.011 0.004 0.004 0.009 0.010 0.012 0.011
-t 79.235 4.973 1.536 2.187 4.133 96.004 74.929 4.335
-Class2 0.075 0.105 0.819 0.676 0.000 0.324 0.087 0.090
-se 0.084 NaN 0.072 0.317 0.387 0.180 0.084 0.103
-t 0.893 NA 11.377 2.133 0.000 1.802 1.037 0.877
-Class3 0.036 0.960 0.004 0.000 0.939 0.061 0.026 0.935
-se 0.015 0.016 0.005 NaN 0.007 0.020 0.013 0.017
-t 2.377 59.545 0.775 NA 133.115 3.016 2.057 55.569
- Item3,p 3 Item4,p 1 Item4,p 2 Item4,p 3
-Class1 0.017 0.004 0.026 0.970
-se 0.006 0.003 0.007 0.008
-t 2.873 1.393 3.458 121.763
-Class2 0.823 0.733 0.173 0.094
-se 0.123 0.152 0.054 0.142
-t 6.719 4.833 3.198 0.667
-Class3 0.039 0.004 0.930 0.065
-se 0.012 0.004 0.022 0.022
-t 3.337 1.001 41.420 2.959
-
-
- Parameter values, unconditional (class) probabilities
-
- Class1 Class2 Class3
-val 0.689 0.017 0.294
-se 0.018 0.005 0.017
-t 39.127 3.331 16.941
-
-Model: 4 class model fitted at Mon Feb 18 11:59:58 2008
-Optimization information, method is donlp
- Iterations: 80
- Inform: KT-conditions (relaxed) satisfied, singular point (look up the respective manuals for more information.)
-
- Loglikelihood of fitted model: -1220.541
- AIC: 2511.081
- BIC: 2674.112
- Number of observations (used in BIC): 779
- Fitted model
- Model: 4 class model
- Number of parameters: 69
- Free parameters: 35
- Number of states: 4
- Number of items: 4
- Item types: 3 3 3 3
-
- Parameter values, observation parameters
-
- Item1,p 1 Item1,p 2 Item1,p 3 Item2,p 1 Item2,p 2 Item2,p 3 Item3,p 1 Item3,p 2
-Class1 0.978 0.020 0.002 0.007 0.018 0.975 0.981 0.011
-se 0.032 0.027 0.006 0.005 0.014 0.016 0.037 0.028
-t 30.972 0.765 0.338 1.325 1.328 59.916 26.512 0.410
-Class2 0.022 0.978 0.000 0.000 0.972 0.028 0.017 0.953
-se 0.013 0.072 0.073 NaN 0.018 0.020 0.011 0.018
-t 1.651 13.651 0.000 NA 54.756 1.437 1.493 52.043
-Class3 0.087 0.079 0.834 0.744 0.000 0.256 0.087 0.085
-se 0.089 NaN 0.043 0.245 0.261 0.131 0.090 0.084
-t 0.980 NA 19.473 3.038 0.000 1.953 0.961 1.012
-Class4 0.499 0.448 0.053 0.025 0.283 0.692 0.440 0.441
-se 0.295 0.307 0.040 0.028 0.248 0.249 0.315 0.294
-t 1.690 1.458 1.335 0.878 1.143 2.782 1.396 1.502
- Item3,p 3 Item4,p 1 Item4,p 2 Item4,p 3
-Class1 0.007 0.004 0.017 0.979
-se 0.011 0.003 0.010 0.010
-t 0.692 1.412 1.695 98.374
-Class2 0.030 0.005 0.977 0.018
-se 0.013 0.005 0.028 0.028
-t 2.359 1.022 34.440 0.641
-Class3 0.828 0.836 0.164 0.000
-se 0.118 0.268 0.060 0.273
-t 7.012 3.126 2.707 0.000
-Class4 0.119 0.000 0.177 0.823
-se 0.050 NaN 0.182 0.194
-t 2.354 NA 0.973 4.250
-
-
- Parameter values, unconditional (class) probabilities
-
- Class1 Class2 Class3 Class4
-val 0.616 0.270 0.015 0.099
-se 0.088 0.023 0.004 0.074
-t 7.001 11.687 3.667 1.331
-
-
-
-#
-# 3-class model is best according to BIC
-#
-
-# get the posteriors from it
-
-dat$post <- as.factor(fit3$post$states[[1]][,2])
-
-levels(dat$post) <- c("rule2","trans","rule1")
-
-> lm(ageyears~post,dat=dat)
-
-Call:
-lm(formula = ageyears ~ post, data = dat)
-
-Coefficients:
-(Intercept) posttrans postrule1
- 12.7908 -0.3293 -3.6279
-
-> plot(ageyears~post,dat=dat)
-> summary(multinom(post~ageyears,dat=dat))
-# weights: 9 (4 variable)
-initial value 855.818973
-iter 10 value 398.743386
-final value 393.199188
-converged
-Call:
-multinom(formula = post ~ ageyears, data = dat)
-
-Coefficients:
- (Intercept) ageyears
-trans -3.190456 -0.04319348
-rule1 5.233625 -0.56814085
-
-Std. Errors:
- (Intercept) ageyears
-trans 1.2964214 0.10150279
-rule1 0.4683135 0.04524304
-
-Residual Deviance: 786.3984
-AIC: 794.3984
-
-
-
-
-
-
-#
-# DICHOTOME DATA MODELLEN
-#
-
-
-setwd("/Users/ivisser/Documents/projects/depmixProject/lcavoorbeeld/")
-
-dat=read.table("lca.txt",head=T)
-
-# trichotome data
-dat <- dat[,c("d2_di","d3_bi","d4_di","d5_di","ageyears")]
-
-Model: 1 class model fitted at Fri Jan 25 14:48:04 2008
-Optimization information, method is donlp
- Iterations: 10
- Inform: KT-conditions satisfied, no further correction computed (look up the respective manuals for more information.)
-
- Loglikelihood of fitted model: -1973.626
- AIC: 3955.253
- BIC: 3973.885
- Number of observations (used in BIC): 779
- Fitted model
- Model: 1 class model
- Number of parameters: 11
- Free parameters: 4
- Number of states: 1
- Number of items: 4
- Item types: 2 2 2 2
-
- Parameter values, observation parameters
-
- Item1,p 1 Item1,p 2 Item2,p 1 Item2,p 2 Item3,p 1 Item3,p 2 Item4,p 1 Item4,p 2
-Class1 0.341 0.659 0.320 0.680 0.347 0.653 0.311 0.689
-se 0.017 0.017 0.017 0.017 0.017 0.017 0.017 0.017
-t 20.098 38.761 19.131 40.720 20.328 38.322 18.737 41.576
-
-
- Parameter values, unconditional (class) probabilities
-
- Class1
-val 1
-se 0
-t NA
-
-
-
-
-Model: 2 class model fitted at Fri Jan 25 14:47:02 2008
-Optimization information, method is donlp
- Iterations: 18
- Inform: KT-conditions satisfied, no further correction computed (look up the respective manuals for more information.)
-
- Loglikelihood of fitted model: -1083.036
- AIC: 2184.073
- BIC: 2225.995
- Number of observations (used in BIC): 779
- Fitted model
- Model: 2 class model
- Number of parameters: 23
- Free parameters: 9
- Number of states: 2
- Number of items: 4
- Item types: 2 2 2 2
-
- Parameter values, observation parameters
-
- Item1,p 1 Item1,p 2 Item2,p 1 Item2,p 2 Item3,p 1 Item3,p 2 Item4,p 1 Item4,p 2
-Class1 0.965 0.035 0.929 0.071 0.970 0.030 0.934 0.066
-se 0.015 0.015 0.021 0.021 0.013 0.013 0.022 0.022
-t 64.440 2.314 43.618 3.350 74.558 2.270 42.164 2.982
-Class2 0.062 0.938 0.047 0.953 0.067 0.933 0.031 0.969
-se 0.012 0.012 0.010 0.010 0.013 0.013 0.008 0.008
-t 5.123 77.910 4.680 95.892 5.220 72.868 3.814 118.626
-
-
- Parameter values, unconditional (class) probabilities
-
- Class1 Class2
-val 0.310 0.690
-se 0.018 0.018
-t 17.449 38.911
-
-
-Model: 3 class model fitted at Fri Jan 25 14:49:39 2008
-Optimization information, method is donlp
- Iterations: 27
- Inform: KT-conditions (relaxed) satisfied, singular point (look up the respective manuals for more information.)
-
- Loglikelihood of fitted model: -1062.374
- AIC: 2152.747
- BIC: 2217.96
- Number of observations (used in BIC): 779
- Fitted model
- Model: 3 class model
- Number of parameters: 37
- Free parameters: 14
- Number of states: 3
- Number of items: 4
- Item types: 2 2 2 2
-
- Parameter values, observation parameters
-
- Item1,p 1 Item1,p 2 Item2,p 1 Item2,p 2 Item3,p 1 Item3,p 2 Item4,p 1 Item4,p 2
-Class1 0.982 0.018 0.962 0.038 0.981 0.019 0.982 0.018
-se 0.038 0.038 0.071 0.071 0.027 0.027 0.137 0.137
-t 26.003 0.471 13.548 0.529 36.056 0.699 7.180 0.128
-Class2 0.023 0.977 0.024 0.976 0.017 0.983 0.019 0.981
-se 0.109 0.109 0.055 0.055 0.159 0.159 0.030 0.030
-t 0.211 8.943 0.438 17.879 0.109 6.168 0.647 32.936
-Class3 0.473 0.527 0.314 0.686 0.560 0.440 0.201 0.799
-se 1.376 1.376 1.155 1.155 1.365 1.365 0.854 0.854
-t 0.344 0.383 0.272 0.594 0.410 0.322 0.236 0.935
-
-
- Parameter values, unconditional (class) probabilities
-
- Class1 Class2 Class3
-val 0.283 0.613 0.105
-se 0.099 0.381 0.284
-t 2.864 1.609 0.369
-
-# posterior classes related to age in years
-# weights: 9 (4 variable)
-initial value 855.818973
-iter 10 value 526.053934
-final value 526.041488
-converged
-Call:
-multinom(formula = post ~ ageyears, data = dat)
-
-Coefficients:
- (Intercept) ageyears
-2 -5.320429 0.5578123
-3 -4.356987 0.3028828
-
-Std. Errors:
- (Intercept) ageyears
-2 0.4641045 0.04387059
-3 0.6461756 0.05985131
-
-Residual Deviance: 1052.083
-AIC: 1060.083
-
-
-
-Model: 4 class model fitted at Fri Jan 25 14:53:45 2008
-Optimization information, method is donlp
- Iterations: 38
- Inform: KT-conditions (relaxed) satisfied, singular point (look up the respective manuals for more information.)
-
- Loglikelihood of fitted model: -1058.019
- AIC: 2154.038
- BIC: 2242.54
- Number of observations (used in BIC): 779
- Fitted model
- Model: 4 class model
- Number of parameters: 53
- Free parameters: 19
- Number of states: 4
- Number of items: 4
- Item types: 2 2 2 2
-
- Parameter values, observation parameters
-
- Item1,p 1 Item1,p 2 Item2,p 1 Item2,p 2 Item3,p 1 Item3,p 2 Item4,p 1 Item4,p 2
-Class1 0.995 0.005 0.987 0.013 1.000 0.000 0.978 0.022
-se 0.042 0.042 NaN NaN NaN NaN 0.078 0.078
-t 23.685 0.121 NA NA NA NA 12.507 0.278
-Class2 0.700 0.300 0.000 1.000 1.000 0.000 0.388 0.612
-se 0.063 0.063 2.732 2.732 NaN NaN NaN NaN
-t 11.158 4.786 0.000 0.366 NA NA NA NA
-Class3 0.567 0.433 1.000 0.000 0.511 0.489 0.527 0.473
-se NaN NaN 0.029 0.029 1.122 1.122 NaN NaN
-t NA NA 34.293 0.000 0.456 0.436 NA NA
-Class4 0.042 0.958 0.025 0.975 0.033 0.967 0.023 0.977
-se 0.009 0.009 0.009 0.009 0.011 0.011 0.007 0.007
-t 4.632 106.510 2.722 105.363 2.894 85.135 3.372 143.828
-
-
- Parameter values, unconditional (class) probabilities
-
- Class1 Class2 Class3 Class4
-val 0.263 0.040 0.044 0.654
-se 0.016 0.109 0.101 0.016
-t 16.785 0.365 0.433 41.634
-
-
-
-
-
-
More information about the depmix-commits
mailing list