BIOdry (version 0.3)

ringLme: LME modeling

Description

LME modeling of multilevel data frames.

Usage

ringLme(rd, form = "lmeForm", var.mod = FALSE, arma.mod = FALSE,  res.data = TRUE, ...)

Arguments

rd

data.frame object containing factor-level columns.

form

character. Method name of LME formula. Available methods are 'lmeForm' and 'tdForm' (see details).

var.mod

logical. If TRUE then the fitted model is extended by modeling residual heteroscedaticity with 'varConstPower' class in nlme.

arma.mod

logical. If TRUE then the fitted model is extended by modeling residual autocorrelation with class 'corARMA' in nlme, with p = 1 and q = 1, which models residual autocorrelation for lags > 2.

res.data

logical. Save residuals as a multilevel data frame. If TRUE then a data frame of name 'resid' is added to output list

...

Further arguments to be passed to function in form.

Value

res.data, either an LME model, or a list with both: the LME model, and a residual data frame containing initial factor-level columns.

Details

This function implements lme to fit linear mixed-effects models on data frames with factor-level columns. Two kind of model formulas are available: 'lmeForm' and 'tdForm', these characters implement functions with same names (tdForm and lmeForm). Users can develop their own LME formulas by modifying arguments in any of these methods. After LME models are fitted, they can be extended by modeling heteroscedasticity and autocorrelation of the residuals. Nevertheless, such residual modeling would take long time depending on the complexity of the modeled levels.

References

Pinheiro J. C., D. M. Bates. 2000. Mixed-effects models in S and S-PLUS. Springer, New York.

Examples

##Multilevel data frame of tree-ring widths:
data(Prings05,envir = environment())
## Radial increments measured on 2003:
data(Pradii03,envir = environment())    
## Monthly precipitation sums and average temperatures:
data(PTclim05,envir = environment())

##Modeling tree growth:
mpin <- modelFrame(Prings05, y = Pradii03,
        form = NULL,# on.time = TRUE,
        MoreArgs = list(only.dup = TRUE,
        mp = c(1,1),un = c('mm','cm'),z = 2003))

## Detrending tree growth with a td-form model:
rlme <- ringLme(mpin,form = 'tdForm')
summary(rlme$model)

##a plot of the modeled fluctuations
d <- groupedData(lmeForm(rlme$resid,lev.rm = 1),data = rlme$resid)
plot(d,groups = ~ sample,auto.key = TRUE)

## A model of aridity: 
cf <- modelFrame(PTclim05,
     lv = list('year','year'),
     fn = list('moveYr','wlai'),
     form = NULL)
summary(cf)

## An lme model of aridity at 'plot' level:
rmod <- ringLme(cf,form = 'lmeForm')
summary(rmod$model)

rk <- groupedData(lmeForm(rmod$resid),data=rmod$resid)
plot(rk,ylab = 'detrended AI')