thr3ads.net - R help - [R] Help: lme [Jun 2006]

If this information is useful, please help other people find it:
Share via:

Pryseley Assam

2006-Jun-01 10:25 UTC

[R] Help: lme

Good day R-Users,
   
  I have a problem accessing some values in the output from the summary of an
lme fit.
   
  
The structure of my data is as shown below (I have attached a copy of the full
data).
  
 
  id trials endp Z.sas       ST
  1      1   -1         -1        42.42884
  1      1    1         -1         48.12007
  2      1   -1         -1        43.42878
  2      1    1         -1         46.82817
  3      1   -1         -1        45.56736
  













.
  598  10  -1        1          49.59715
  598  10   1        1          55.46324
  599  10   -1       1          46.80873
  599  10    1       1          53.52223
  600  10   -1       1          48.58257
  600  10    1       1          56.78674
  
 I used the following code to fit my model of interest:
   
  ggg <- lme (ST~ -1 + as.factor(endp):Z.sas + as.factor(endp), data=dat4a,
  random=~-1 + as.factor(endp) + as.factor(endp):Z.sas|as.factor(trials),
  correlation = corSymm(form=~1|as.factor(trials)/as.factor(id)),
weights=varIdent(form=~1|endp))
   
  hh <- summary(ggg)
  hh
   
  Below is the following part of the output of interest I wish to access:
   
  Correlation Structure: General
 Formula: ~1 | as.factor(trials)/as.factor(id) 
 Parameter estimate(s):
 Correlation: 
  1    
2 0.785
Variance function:
 Structure: Different standard deviations per stratum
 Formula: ~1 | endp 
 Parameter estimates:
       -1         1 
1.0000000 0.9692405 
   
  I wish to access the value of the correlation (0.785) and the vector of the
variance function estimates (1, 0.969). I know these can be done through the
intervals function, but sometimes when the estimated Hessian matrix is not
positive definite or something like that (i am not quite sure), the intervals
function delivers an error message.
   
  Thus, i will like to ask if there is another way to access these values. I
tried using the following code:
   
  hh$modelStruct$corStruct[1]
  hh$modelStruct$varStruct[1]
   
  Rather the output was:
   
  > hh$modelStruct$corStruct[1]
[1] -1.308580> hh$modelStruct$varStruct[1][1] -0.03124255
   
  I presume there is a way to calculate the correlation and variance function
coefficients using these values.
   
  Could you tell me how to access those values (without using the intervals
function) or better still how to calculate the values from the last output
values.
   
  Kind regards
  Pryseley
   
   

__________________________________________________


-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: dat4a.txt
Url:
https://stat.ethz.ch/pipermail/r-help/attachments/20060601/49b07c35/attachment.txt

Doran, Harold

2006-Jun-01 11:37 UTC

head link

[R] Help: lme

Pryseley:

You asked this question last week and received answers to that question
along with other suggestions from Doug Bates and myself. Is there some
reason the post by Andrew Robinson is not working?

Harold
> -----Original Message-----
> From: r-help-bounces at stat.math.ethz.ch 
> [mailto:r-help-bounces at stat.math.ethz.ch] On Behalf Of Pryseley Assam
> Sent: Thursday, June 01, 2006 6:25 AM
> To: r-help at stat.math.ethz.ch
> Subject: [R] Help: lme
> 
> Good day R-Users,
>    
>   I have a problem accessing some values in the output from 
> the summary of an lme fit.
>    
>   
> The structure of my data is as shown below (I have attached a 
> copy of the full data).
>   
>  
>   id trials endp Z.sas       ST
>   1      1   -1         -1        42.42884
>   1      1    1         -1         48.12007
>   2      1   -1         -1        43.42878
>   2      1    1         -1         46.82817
>   3      1   -1         -1        45.56736
>   ...........................................
>   598  10  -1        1          49.59715
>   598  10   1        1          55.46324
>   599  10   -1       1          46.80873
>   599  10    1       1          53.52223
>   600  10   -1       1          48.58257
>   600  10    1       1          56.78674
>   
>  I used the following code to fit my model of interest:
>    
>   ggg <- lme (ST~ -1 + as.factor(endp):Z.sas + 
> as.factor(endp), data=dat4a,
>   random=~-1 + as.factor(endp) + 
> as.factor(endp):Z.sas|as.factor(trials),
>   correlation = 
> corSymm(form=~1|as.factor(trials)/as.factor(id)), 
> weights=varIdent(form=~1|endp))
>    
>   hh <- summary(ggg)
>   hh
>    
>   Below is the following part of the output of interest I 
> wish to access:
>    
>   Correlation Structure: General
>  Formula: ~1 | as.factor(trials)/as.factor(id)  Parameter estimate(s):
>  Correlation: 
>   1    
> 2 0.785
> Variance function:
>  Structure: Different standard deviations per stratum
>  Formula: ~1 | endp
>  Parameter estimates:
>        -1         1 
> 1.0000000 0.9692405 
>    
>   I wish to access the value of the correlation (0.785) and 
> the vector of the variance function estimates (1, 0.969). I 
> know these can be done through the intervals function, but 
> sometimes when the estimated Hessian matrix is not positive 
> definite or something like that (i am not quite sure), the 
> intervals function delivers an error message. 
>    
>   Thus, i will like to ask if there is another way to access 
> these values. I tried using the following code:
>    
>   hh$modelStruct$corStruct[1]
>   hh$modelStruct$varStruct[1]
>    
>   Rather the output was:
>    
>   > hh$modelStruct$corStruct[1]
> [1] -1.308580
> > hh$modelStruct$varStruct[1]
> [1] -0.03124255
>    
>   I presume there is a way to calculate the correlation and 
> variance function coefficients using these values. 
>    
>   Could you tell me how to access those values (without using 
> the intervals function) or better still how to calculate the 
> values from the last output values. 
>    
>   Kind regards
>   Pryseley
>    
>    
> 
> __________________________________________________
> 
> 
>

Doran, Harold

2006-Jun-01 13:48 UTC

head link

[R] Help: lme

OK, in the future it might be helpful to note what you tried and what
error messages you received when you worked with the syntax provided.
Why did VarCorr() not work?
 
 In general, there seems to be a larger probelm with your model as Doug
Bates noted. I would also suggest that you go about this another way.
Instead of using lme(), switch to lmer and then use the MCMCsamp()
function to investigate the parameters of the model. To see what can be
accesed from the fitted lmer object use str(model) or use the other
extractor functions such as VarCorr when needed.


________________________________

	From: Pryseley Assam [mailto:assampryseley@yahoo.com] 
	Sent: Thursday, June 01, 2006 9:38 AM
	To: Doran, Harold
	Cc: r-help@stat.math.ethz.ch
	Subject: RE: [R] Help: lme
	
	
	Good day,
	 
	Yeah i tried most of the suggestions and still had the same
problem. Some people also suggested that i supplied a data set so they
could verify what i am saying. That is why i am still asking the same
question with more details as required. 
	 
	My main focus now is to calculate the values in the
"unaccessible"-printed output from the accessible ones as mentioned
towards the end of my latest mail.
	 
	Kind regards
	Pryseley
	
	"Doran, Harold" <HDoran@air.org> wrote:

		Pryseley:
		
		You asked this question last week and received answers
to that question
		along with other suggestions from Doug Bates and myself.
Is there some
		reason the post by Andrew Robinson is not working?
		
		Harold
		
		> -----Original Message-----
		> From: r-help-bounces@stat.math.ethz.ch 
		> [mailto:r-help-bounces@stat.math.ethz.ch] On Behalf Of
Pryseley Assam
		> Sent: Thursday, June 01, 2006 6:25 AM
		> To: r-help@stat.math.ethz.ch
		> Subject: [R] Help: lme
		> 
		> Good day R-Users,
		> 
		> I have a problem accessing some values in the output
from 
		> the summary of an lme fit.
		> 
		> 
		> The structure of my data is as shown below (I have
attached a 
		> copy of the full data).
		> 
		> 
		> id trials endp Z.sas ST
		> 1 1 -1 -1 42.42884
		> 1 1 1 -1 48.12007
		> 2 1 -1 -1 43.42878
		> 2 1 1 -1 46.82817
		> 3 1 -1 -1 45.56736
		> ...........................................
		> 598 10 -1 1 49.59715
		> 598 10 1 1 55.46324
		> 599 10 -1 1 46.80873
		> 599 10 1 1 53.52223
		> 600 10 -1 1 48.58257
		> 600 10 1 1 56.78674
		> 
		> I used the following code to fit my model of interest:
		> 
		> ggg <- lme (ST~ -1 + as.factor(endp):Z.sas + 
		> as.factor(endp), data=dat4a,
		> random=~-1 + as.factor(endp) + 
		> as.factor(endp):Z.sas|as.factor(trials),
		> correlation = 
		> corSymm(form=~1|as.factor(trials)/as.factor(id)), 
		> weights=varIdent(form=~1|endp))
		> 
		> hh <- summary(ggg)
		> hh
		> 
		> Below is the following part of the output of interest
I 
		> wish to access:
		> 
		> Correlation Structure: General
		> Formula: ~1 | as.factor(trials)/as.factor(id)
Parameter estimate(s):
		> Correlation: 
		> 1 
		> 2 0.785
		> Variance function:
		> Structure: Different standard deviations per stratum
		> Formula: ~1 | endp
		> Parameter estimates:
		> -1 1 
		> 1.0000000 0.9692405 
		> 
		> I wish to access the value of the correlation (0.785)
and 
		> the vector of the variance function estimates (1,
0.969). I 
		> know these can be done through the intervals function,
but 
		> sometimes when the estimated Hessian matrix is not
positive 
		> definite or something like that (i am not quite sure),
the 
		> intervals function delivers an error message. 
		> 
		> Thus, i will like to ask if there is another way to
access 
		> these values. I tried using the following code:
		> 
		> hh$modelStruct$corStruct[1]
		> hh$modelStruct$varStruct[1]
		> 
		> Rather the output was:
		> 
		> > hh$modelStruct$corStruct[1]
		> [1] -1.308580
		> > hh$modelStruct$varStruct[1]
		> [1] -0.03124255
		> 
		> I presume there is a way to calculate the correlation
and 
		> variance function coefficients using these values. 
		> 
		> Could you tell me how to access those values (without
using 
		> the intervals function) or better still how to
calculate the 
		> values from the last output values. 
		> 
		> Kind regards
		> Pryseley
		> 
		> 
		> 
		> __________________________________________________
		> 
		> 
		> 
		


	__________________________________________________
	Do You Yahoo!?
	Tired of spam? Yahoo! Mail has the best spam protection around 
	http://mail.yahoo.com 


	[[alternative HTML version deleted]]

Seemingly Similar Threads

Search for more reasonably related threads

R help - Jun 2006 - Help: lme

[R] Help: lme

[R] Help: lme

[R] Help: lme

Seemingly Similar Threads