Skip to content

Multiple covariates in SSbiexp model and an error message

1 message · Andrew Close

Original
Andrew Close
# 
Dear all,

I wanted to experiment with the SSbiexp model but include multiple covariates. To this end I have been following the presented in P&B for the CO2 uptake model

I was presented with the error message:

### Error in eval(expr, envir, enclos) : object 'temp1' not found ###

The following code illustrates my modelling process:
I attached a new column to the Indometh data from randomly generated numbers to represent temperature.
library(nlme)
temp<-rnorm(66, mean = 3, sd = 0.5)
temp<-data.frame(temp)
temp
Indometh1<-cbind(Indometh, temp)

I converted the temperature values to the same order of magnitude using the following line of code.
converted the data to a grouped object and began the model fitting process:
Nonlinear mixed-effects model fit by maximum likelihood
  Model: conc ~ SSbiexp(time + H, A1, lrc1, A2, lrc2)
 Data: Indometh1.g
        AIC       BIC   logLik
  -91.35715 -71.65026 54.67858
Random effects:
 Formula: list(A1 ~ 1, lrc1 ~ 1, A2 ~ 1, lrc2 ~ 1)
 Level: Subject
 Structure: Diagonal
               A1      lrc1        A2         lrc2  Residual
StdDev: 0.5640351 0.1556456 0.1122092 9.439076e-06 0.0815653
Fixed effects: list(A1 ~ 1, lrc1 ~ 1, A2 ~ 1, lrc2 ~ 1)
          Value Std.Error DF   t-value p-value
A1    2.8328430 0.2613774 57 10.838132   0e+00
lrc1  0.7731807 0.1095290 57  7.059140   0e+00
A2    0.4625497 0.1132419 57  4.084615   1e-04
lrc2 -1.3422416 0.2309100 57 -5.812833   0e+00
 Correlation:
     A1     lrc1   A2
lrc1  0.057
A2   -0.102  0.634
lrc2 -0.140  0.581  0.834
Standardized Within-Group Residuals:
       Min         Q1        Med         Q3        Max
-3.1672960 -0.3592548 -0.1301288  0.3445645  2.9932219
Number of Observations: 66
Number of Groups: 6

##
Using the estimates from the above model I tried to generate new starting values for the fixed effects. My aim to see how the decay may vary according to time and the newly generated "temp1" variable.
+ start=c(2.8328430, 0, 0, 0.7731807, 0, 0.4625497, 0, 0, -1.3422416, 0))

I was presented with the the error message:

### Error in eval(expr, envir, enclos) : object 'temp1' not found ###

Can anyone clarify if it possible to generate biexp models with multiple covariates and if so what is it that I am doing wrong?

Thank you for your patience.

Best wishes

Andrew Close
<mailto:a.j.close at ncl.ac.uk>