# saving confidence intervals to a separate file

5 posts / 0 new
Offline
Joined: 02/21/2019 - 18:37
saving confidence intervals to a separate file

Hello, in asimulation loop I want o to save results for instance confidence intervals for each run in separate file. Hos to do this is rather easy in eg.g. lavaan buit I have trouibles as newbie to find out how to specify the matrix cells for the parameter values i want to save. An example from Jaks book on Meta SEM:

pathmodel= mxModel(title,obs,matrixB,matrixP,matrixI, Ind_pos,Ind_neg,conf,algebraS,exp,fit)
pathmodelOut = mxRun(pathmodel, intervals = TRUE)

# retrieve the output
summary(pathmodelOut) 

results in (among other output) this;

confidence intervals:
lbound    estimate      ubound note
b31                      0.39939069  0.64232806  0.88527142
b32                     -0.50352850 -0.30476191 -0.10599531
b43                      0.12624700  0.29585799  0.46546898


How can I specify these estimates as e.g. matric cells and save in a separate file?

Offline
Joined: 01/24/2014 - 12:15
You can save the model

You can save the model summary as a new object,

pathmodelSumm <- summary(pathmodelOut)

, and extract the cells you want from the element of pathmodelSumm that stores the confidence-interval table.

Offline
Joined: 02/21/2019 - 18:37
saving confidence intervals to a separate file

Thanks for your reply, but that far I understand how to do. The problem for me is to go from this and find out how to define which elements contain the confidence interval values. The first part uf the summary looks like this:
Summary of Path model

free parameters:
name matrix row col Estimate Std.Error A
1 b31 B 3 1 0.6423280 0.12280667
2 b32 B 3 2 -0.3047620 0.10047876
3 b43 B 4 3 0.2958579 0.08574016
4 Path model.P[1,1] P 1 1 0.8022115 0.11124744
5 Path model.P[1,2] P 1 2 -0.3565385 0.10230405
6 Path model.P[2,2] P 2 2 1.1983654 0.16618284
7 Path model.P[3,3] P 3 3 1.0918754 0.15141626
8 Path model.P[4,4] P 4 4 1.2796472 0.17745574

confidence intervals:
lbound estimate ubound note
b31 0.39887350 0.64232796 0.88578253
b32 -0.50387525 -0.30476198 -0.10564865
b43 0.12599465 0.29585794 0.46572124
Path model.P[1,1] 0.61833179 0.80221150 1.06727282
Path model.P[1,2] -0.58940484 -0.35653849 -0.17279121
Path model.P[2,2] 0.92353384 1.19836537 1.59494917
Path model.P[3,3] 0.84150214 1.09187541 1.45306408
Path model.P[4,4] 0.98614251 1.27964720 1.70326762

and I would like to specify, similar to if it was a matrix with values for rows and columns, or sometyhing like that, so I could extraxt just the lines with the confidence interval for b31, b32 and b43. But I do not see how to get information on how to define these elements, or "cells", or "rows" ( if you could treat it as a matrix). Maybe this is elementary, I apologize for me being in the beginning of the learning curve,

Offline
Joined: 05/24/2012 - 00:35
model introspection

The way to figure it out is to use str(model$output) Does this help? Offline Joined: 02/21/2019 - 18:37 saving confidence intervals to a separate file - found out how t After some studying I found out that the steps are 1) rename the summary tp spomething simpler; summodel 2) use names(summodel) to see the objects 3) explore which object has the confidence intervals: it is$CI
4) extract the part I want to use by writing

summodel\$CI[c(1:3),c(1:3)]