Covariance structures with glmmTMB

This vignette demonstrates some of the covariance structures available in the glmmTMB package. Currently the available covariance structures are:

Covariance	Notation	Parameter count	Requirement
Heterogeneous unstructured	`us`	\(n(n+1)/2\)
Heterogeneous Toeplitz	`toep`	\(2n-1\)
Heterogeneous compound symmetry	`cs`	\(n+1\)
Heterogeneous diagonal	`diag`	\(n\)
AR(1)	`ar1`	\(2\)
Ornstein–Uhlenbeck	`ou`	\(2\)	Coordinates
Spatial exponential	`exp`	\(2\)	Coordinates
Spatial Gaussian	`gau`	\(2\)	Coordinates
Spatial Matern	`mat`	\(3\)	Coordinates

The word ‘heterogeneous’ refers to the marginal variances of the model. Beyond correlation parameters, a heterogeneous structure uses \(n\) additional variance parameters where \(n\) is the dimension.

Some of the structures require temporal or spatial coordinates. We will show examples of this in a later section.

The AR(1) covariance structure

Demonstration on simulated data

First, let’s consider a simple time series model. Assume that our measurements \(Y(t)\) are given at discrete times \(t \in \{1,...,n\}\) by

\[Y(t) = \mu + X(t) + \varepsilon(t)\]

where

\(\mu\) is the mean value parameter.
\(X(t)\) is a stationary AR(1) process, i.e. has covariance \(cov(X(s), X(t)) = \sigma^2\exp(-\theta |t-s|)\).
\(\varepsilon(t)\) is iid. \(N(0,\sigma_0^2)\) measurement error.

A simulation experiment is set up using the parameters

Description	Parameter	Value
Mean	\(\mu\)	0
Process variance	\(\sigma^2\)	1
Measurement variance	\(\sigma_0^2\)	1
One-step correlation	\(e^{-\theta}\)	0.7

The following R-code draws a simulation based on these parameter values. For illustration purposes we consider a very short time series.

n <- 6                                              ## Number of time points
x <- mvrnorm(mu = rep(0,n),
             Sigma = .7 ^ as.matrix(dist(1:n)) )    ## Simulate the process using the MASS package
y <- x + rnorm(n)                                   ## Add measurement noise

In order to fit the model with glmmTMB we must first specify a time variable as a factor. The factor levels correspond to unit spaced time points.

times <- factor(1:n)
levels(times)

## [1] "1" "2" "3" "4" "5" "6"

We also need a grouping variable. In the current case there is only one time-series so the grouping is:

group <- factor(rep(1,n))

We combine the data into a single data frame (not absolutely required, but good practice):

dat0 <- data.frame(y,times,group)

Now fit the model using

glmmTMB(y ~ ar1(times + 0 | group), data=dat0)

This formula notation follows that of the lme4 package.

The left hand side of the bar times + 0 corresponds to a design matrix \(Z\) linking observation vector \(y\) (rows) with a random effects vector \(u\) (columns).
The distribution of \(u\) is ar1 (this is the only glmmTMB specific part of the formula).
The right hand side of the bar splits the above specification independently among groups. Each group has its own separate \(u\) vector but shares the same parameters for the covariance structure.

After running the model, we find the parameter estimates \(\mu\) (intercept), \(\sigma_0^2\) (dispersion), \(\sigma\) (Std. Dev.) and \(e^{-\theta}\) (First off-diagonal of “Corr”) in the output:

FIXME: Try a longer time series when the print.VarCorr is fixed.

## Formula:          y ~ ar1(times + 0 | group)
## Data: dat0
##      AIC      BIC   logLik df.resid 
## 18.76343 17.93047 -5.38171        2 
## Random-effects (co)variances:
## 
## Conditional model:
##  Groups   Name   Std.Dev.  Corr       
##  group    times1 7.018e-01 -0.58 (ar1)
##  Residual        1.165e-05            
## 
## Number of obs: 6 / Conditional model: group, 1
## 
## Dispersion estimate for gaussian family (sigma^2): 1.36e-10 
## 
## Fixed Effects:
## 
## Conditional model:
## (Intercept)  
##       1.072

Increasing the sample size

A single time series of 6 time points is not sufficient to identify the parameters. We could either increase the length of the time series or increase the number of groups. We’ll try the latter:

simGroup <- function(g) {
    x <- mvrnorm(mu = rep(0,n),
             Sigma = .7 ^ as.matrix(dist(1:n)) )    ## Simulate the process
    y <- x + rnorm(n)                               ## Add measurement noise
    times <- factor(1:n)
    group <- factor(rep(g,n))
    data.frame(y, times, group)
}
simGroup(1)

##           y times group
## 1 0.3008231     1     1
## 2 0.1835385     2     1
## 3 0.3750261     3     1
## 4 1.5819573     4     1
## 5 1.5776466     5     1
## 6 0.2072020     6     1

Generate a dataset with 1000 groups:

dat1 <- do.call("rbind", lapply(1:1000, simGroup) )

And fitting the model on this larger dataset gives estimates close to the true values:

(fit.ar1 <- glmmTMB(y ~ ar1(times + 0 | group), data=dat1))

## Formula:          y ~ ar1(times + 0 | group)
## Data: dat1
##       AIC       BIC    logLik  df.resid 
##  20549.90  20576.70 -10270.95      5996 
## Random-effects (co)variances:
## 
## Conditional model:
##  Groups   Name   Std.Dev. Corr      
##  group    times1 1.003    0.72 (ar1)
##  Residual        1.019              
## 
## Number of obs: 6000 / Conditional model: group, 1000
## 
## Dispersion estimate for gaussian family (sigma^2): 1.04 
## 
## Fixed Effects:
## 
## Conditional model:
## (Intercept)  
##     0.04378

The unstructured covariance

We can try to fit an unstructured covariance to the previous dataset dat. For this case an unstructured covariance has 15 correlation parameters and 6 variance parameters. Adding \(\sigma_0^2 I\) on top would cause a strict overparameterization. Hence, when fitting the model with glmmTMB, we have to disable the \(\varepsilon\) term (the dispersion):

fit.us <- glmmTMB(y ~ us(times + 0 | group), data=dat1, dispformula=~0)
fit.us$sdr$pdHess ## Converged ?

## [1] TRUE

The estimated variance and correlation parameters are:

VarCorr(fit.us)

## 
## Conditional model:
##  Groups   Name   Std.Dev.   Corr                          
##  group    times1 1.41313653                               
##           times2 1.38976885 0.353                         
##           times3 1.50181088 0.273 0.384                   
##           times4 1.40399008 0.166 0.216 0.352             
##           times5 1.43307431 0.160 0.192 0.239 0.354       
##           times6 1.43753488 0.108 0.136 0.195 0.273 0.345 
##  Residual        0.00012207

The estimated correlation is approximately constant along diagonals (apparent Toeplitz structure) and we note that the first off-diagonal is now ca. half the true value (0.7) because the dispersion is effectively included in the estimated covariance matrix.

The Toeplitz structure

The next natural step would be to reduce the number of parameters by collecting correlation parameters within the same off-diagonal. This amounts to 5 correlation parameters and 6 variance parameters.

FIXME: Explain why dispformula=~1 causes over-parameterization

fit.toep <- glmmTMB(y ~ toep(times + 0 | group), data=dat1, dispformula=~0)
fit.toep$sdr$pdHess ## Converged ?

## [1] TRUE

The estimated variance and correlation parameters are:

VarCorr(fit.toep)

## 
## Conditional model:
##  Groups   Name   Std.Dev.   Corr                          
##  group    times1 1.41255279                               
##           times2 1.38784528 0.358                         
##           times3 1.49610367 0.249 0.358                   
##           times4 1.40698827 0.188 0.249 0.358             
##           times5 1.43694789 0.148 0.188 0.249 0.358       
##           times6 1.43875115 0.106 0.148 0.188 0.249 0.358 
##  Residual        0.00012207

The residual variance appears downward biased. REML estimation (currently not part of glmmTMB) would probably give a better estimate of the variance and thereby the correlation parameters.

FIXME: Add REML argument to glmmTMB

Compound symmetry

The compound symmetry structure collects all off-diagonal elements of the correlation matrix to one common value.

FIXME: Explain why dispformula=~1 causes over-parameterization

fit.cs <- glmmTMB(y ~ cs(times + 0 | group), data=dat1, dispformula=~0)
fit.cs$sdr$pdHess ## Converged ?

## [1] TRUE

The estimated variance and correlation parameters are:

VarCorr(fit.cs)

## 
## Conditional model:
##  Groups   Name   Std.Dev.   Corr      
##  group    times1 1.43122953 0.250 (cs)
##  Residual        0.00012207

Anova tables

The models ar1, toep, and us are nested so we can use:

anova(fit.ar1, fit.toep, fit.us)

## Data: dat1
## Models:
## fit.ar1: y ~ ar1(times + 0 | group), zi=~0, disp=~1
## fit.toep: y ~ toep(times + 0 | group), zi=~0, disp=~0
## fit.us: y ~ us(times + 0 | group), zi=~0, disp=~0
##          Df   AIC   BIC logLik deviance  Chisq Chi Df Pr(>Chisq)
## fit.ar1   4 20550 20577 -10271    20542                         
## fit.toep 12 20556 20637 -10266    20532 9.4895      8     0.3027
## fit.us   22 20570 20717 -10263    20526 6.5103     10     0.7707

The model cs is a sub-model of toep:

anova(fit.cs, fit.toep)

## Data: dat1
## Models:
## fit.cs: y ~ cs(times + 0 | group), zi=~0, disp=~0
## fit.toep: y ~ toep(times + 0 | group), zi=~0, disp=~0
##          Df   AIC   BIC logLik deviance  Chisq Chi Df Pr(>Chisq)    
## fit.cs    8 20703 20756 -10343    20687                             
## fit.toep 12 20556 20637 -10266    20532 154.48      4  < 2.2e-16 ***
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1

Adding coordinate information

Coordinate information can be added to a variable using the glmmTMB function numFactor. This is necessary in order to use those covariance structures that require coordinates. For example, if we have the numeric coordinates

x <- sample(1:2, 10, replace=TRUE)
y <- sample(1:2, 10, replace=TRUE)

we can generate a factor representing \((x,y)\) coordinates by

(pos <- numFactor(x,y))

##  [1] (2,1) (2,2) (2,1) (1,2) (2,1) (1,1) (1,2) (2,2) (1,1) (1,2)
## Levels: (1,1) (2,1) (1,2) (2,2)

Numeric coordinates can be recovered from the factor levels:

parseNumLevels(levels(pos))

##      [,1] [,2]
## [1,]    1    1
## [2,]    2    1
## [3,]    1    2
## [4,]    2    2

In order to try the remaining structures on our test data we re-interpret the time factor using numFactor:

dat1$times <- numFactor(dat1$times)
levels(dat1$times)

## [1] "(1)" "(2)" "(3)" "(4)" "(5)" "(6)"

Ornstein–Uhlenbeck

Having the numeric times encoded in the factor levels we can now try the Ornstein–Uhlenbeck covariance structure.

fit.ou <- glmmTMB(y ~ ou(times + 0 | group), data=dat1)
fit.ou$sdr$pdHess ## Converged ?

## [1] TRUE

It should give the exact same results as ar1 in this case since the times are equidistant:

VarCorr(fit.ou)

## 
## Conditional model:
##  Groups   Name     Std.Dev. Corr                          
##  group    times(1) 1.0034                                 
##           times(2) 1.0034   0.721                         
##           times(3) 1.0034   0.520 0.721                   
##           times(4) 1.0034   0.375 0.520 0.721             
##           times(5) 1.0034   0.270 0.375 0.520 0.721       
##           times(6) 1.0034   0.195 0.270 0.375 0.520 0.721 
##  Residual          1.0187

However, note the differences between ou and ar1:

ou can handle irregular time points.
ou only allows positive correlation between neighboring time points.

Spatial correlations

The structures exp, gau and mat are meant to used for spatial data. They all require a Euclidean distance matrix which is calculated internally based on the coordinates. Here, we will try these models on the simulated time series data.

An example with spatial data is presented in a later section.

Matern

fit.mat <- glmmTMB(y ~ mat(times + 0 | group), data=dat1, dispformula=~0)
fit.mat$sdr$pdHess ## Converged ?

## [1] TRUE

VarCorr(fit.mat)

## 
## Conditional model:
##  Groups   Name     Std.Dev.   Corr                          
##  group    times(1) 1.42996896                               
##           times(2) 1.42996896 0.357                         
##           times(3) 1.42996896 0.250 0.357                   
##           times(4) 1.42996896 0.187 0.250 0.357             
##           times(5) 1.42996896 0.143 0.187 0.250 0.357       
##           times(6) 1.42996896 0.111 0.143 0.187 0.250 0.357 
##  Residual          0.00012207

Gaussian

“Gaussian” refers here to a Gaussian decay in correlation with distance, i.e. \(\rho = \exp(-d x^2)\), not to the conditional distribution (“family”).

fit.gau <- glmmTMB(y ~ gau(times + 0 | group), data=dat1, dispformula=~0)
fit.gau$sdr$pdHess ## Converged ?

## [1] TRUE

VarCorr(fit.gau)

## 
## Conditional model:
##  Groups   Name     Std.Dev.   Corr                          
##  group    times(1) 1.41749843                               
##           times(2) 1.41749843 0.280                         
##           times(3) 1.41749843 0.006 0.280                   
##           times(4) 1.41749843 0.000 0.006 0.280             
##           times(5) 1.41749843 0.000 0.000 0.006 0.280       
##           times(6) 1.41749843 0.000 0.000 0.000 0.006 0.280 
##  Residual          0.00012207

Exponential

fit.exp <- glmmTMB(y ~ exp(times + 0 | group), data=dat1)
fit.exp$sdr$pdHess ## Converged ?

## [1] TRUE

VarCorr(fit.exp)

## 
## Conditional model:
##  Groups   Name     Std.Dev. Corr                          
##  group    times(1) 1.0034                                 
##           times(2) 1.0034   0.721                         
##           times(3) 1.0034   0.520 0.721                   
##           times(4) 1.0034   0.375 0.520 0.721             
##           times(5) 1.0034   0.270 0.375 0.520 0.721       
##           times(6) 1.0034   0.195 0.270 0.375 0.520 0.721 
##  Residual          1.0187

A spatial covariance example

Starting out with the built in volcano dataset we reshape it to a data.frame with pixel intensity z and pixel position x and y:

d <- data.frame(z = as.vector(volcano),
                x = as.vector(row(volcano)),
                y = as.vector(col(volcano)))

Next, add random normal noise to the pixel intensities and extract a small subset of 100 pixels. This is our spatial dataset:

set.seed(1)
d$z <- d$z + rnorm(length(volcano), sd=15)
d <- d[sample(nrow(d), 100), ]

Display sampled noisy volcano data:

volcano.data <- array(NA, dim(volcano))
volcano.data[cbind(d$x, d$y)] <- d$z
image(volcano.data, main="Spatial data")

Based on this data, we’ll attempt to re-construct the original image.

As model, it is assumed that the original image image(volcano) is a realization of a random field with correlation decaying exponentially with distance between pixels.

Denoting by \(u(x,y)\) this random field the model for the observations is

\[ z_{i} = \mu + u(x_i,y_i) + \varepsilon_i \]

To fit the model, a numFactor and a dummy grouping variable must be added to the dataset:

d$pos <- numFactor(d$x, d$y)
d$group <- factor(rep(1, nrow(d)))

The model is fit by

f <- glmmTMB(z ~ 1 + exp(pos + 0 | group), data=d)

Recall that a standard deviation sd=15 was used to distort the image. A confidence interval for this parameter is

confint(f, "sigma")

##          2.5 %   97.5 % Estimate
## sigma 9.492257 16.46098 12.50007

The glmmTMB predict method can predict unseen levels of the random effects. For instance to predict a 3-by-3 corner of the image one could construct the new data:

newdata <- data.frame( pos=numFactor(expand.grid(x=1:3,y=1:3)) )
newdata$group <- factor(rep(1, nrow(newdata)))
newdata

##     pos group
## 1 (1,1)     1
## 2 (2,1)     1
## 3 (3,1)     1
## 4 (1,2)     1
## 5 (2,2)     1
## 6 (3,2)     1
## 7 (1,3)     1
## 8 (2,3)     1
## 9 (3,3)     1

and predict using

predict(f, newdata, type="response", allow.new.levels=TRUE)

## [1] 104.4634 104.8042 105.1246 104.6228 104.9824 105.3195 104.7828 105.1666
## [9] 105.5366

A specific image column can thus be predicted using the function

predict_col <- function(i) {
    newdata <- data.frame( pos = numFactor(expand.grid(1:87,i)))
    newdata$group <- factor(rep(1,nrow(newdata)))
    predict(f, newdata=newdata, type="response", allow.new.levels=TRUE)
}

Prediction of the entire image is carried out by (this takes a while…):

pred <- sapply(1:61, predict_col)

Finally plot the re-constructed image by

image(pred, main="Reconstruction")

Mappings

For various advanced purposes, such as computing likelihood profiles, it is useful to know the details of the parameterization of the models - the scale on which the parameters are defined (e.g. standard deviation, variance, or log-standard deviation for variance parameters) and their order.

Unstructured

For an unstructured matrix of size n parameters 1:n represent the log-standard deviations while the remaining n(n-1)/2 (i.e. (n+1):(n:(n*(n+1)/2))) are the elements of the Cholesky factor of the correlation matrix, filled in row-wise order (see TMB documentation)

vv0 <- VarCorr(fit.us)
vv1 <- vv0$cond$group          ## extract 'naked' V-C matrix
n <- nrow(vv1)
rpars <- getME(fit.us,"theta") ## extract V-C parameters
## first n parameters are log-std devs:
all.equal(unname(diag(vv1)),exp(rpars[1:n])^2)

## [1] TRUE

## now try correlation parameters:
cpars <- rpars[-(1:n)]
length(cpars)==n*(n-1)/2      ## the expected number

## [1] TRUE

cc <- diag(n)
cc[upper.tri(cc)] <- cpars
L <- crossprod(cc)
D <- diag(1/sqrt(diag(L)))
D %*% L %*% D

##           [,1]      [,2]      [,3]      [,4]      [,5]      [,6]
## [1,] 1.0000000 0.3534513 0.2727986 0.1655881 0.1600080 0.1076348
## [2,] 0.3534513 1.0000000 0.3840508 0.2164539 0.1920130 0.1363969
## [3,] 0.2727986 0.3840508 1.0000000 0.3519413 0.2394706 0.1949200
## [4,] 0.1655881 0.2164539 0.3519413 1.0000000 0.3543410 0.2731236
## [5,] 0.1600080 0.1920130 0.2394706 0.3543410 1.0000000 0.3445279
## [6,] 0.1076348 0.1363969 0.1949200 0.2731236 0.3445279 1.0000000

unname(attr(vv1,"correlation"))

##           [,1]      [,2]      [,3]      [,4]      [,5]      [,6]
## [1,] 1.0000000 0.3534513 0.2727986 0.1655881 0.1600080 0.1076348
## [2,] 0.3534513 1.0000000 0.3840508 0.2164539 0.1920130 0.1363969
## [3,] 0.2727986 0.3840508 1.0000000 0.3519413 0.2394706 0.1949200
## [4,] 0.1655881 0.2164539 0.3519413 1.0000000 0.3543410 0.2731236
## [5,] 0.1600080 0.1920130 0.2394706 0.3543410 1.0000000 0.3445279
## [6,] 0.1076348 0.1363969 0.1949200 0.2731236 0.3445279 1.0000000

FIXME: why are these not quite the same? Not what I expected

all.equal(c(cov2cor(vv1)),c(fit.us$obj$env$report(fit.us$fit$parfull)$corr[[1]]))

## [1] TRUE

Profiling (experimental/exploratory):

## want $par, not $parfull: do NOT include conditional modes/'b' parameters
ppar <- fit.us$fit$par
length(ppar)

## [1] 22

range(which(names(ppar)=="theta")) ## the last n*(n+1)/2 parameters

## [1]  2 22

## only 1 fixed effect parameter
tt <- tmbprofile(fit.us$obj,2,trace=FALSE)

plot(tt)

confint(tt)

##           lower    upper
## theta 0.3026163 0.390288

ppar <- fit.cs$fit$par
length(ppar)

## [1] 8

range(which(names(ppar)=="theta")) ## the last n*(n+1)/2 parameters

## [1] 2 8

## only 1 fixed effect parameter, 1 dispersion parameter
tt2 <- tmbprofile(fit.cs$obj,3,trace=FALSE)

plot(tt2)

Covariance structures with glmmTMB

Kasper Kristensen

2018-07-02

The AR(1) covariance structure

Demonstration on simulated data

Increasing the sample size

The unstructured covariance

The Toeplitz structure

Compound symmetry

Anova tables

Adding coordinate information

Ornstein–Uhlenbeck

Spatial correlations

Matern

Gaussian

Exponential

A spatial covariance example

Mappings

Unstructured