Diagnostic plots for the linear mixed-effects fit are obtained. The
form argument gives considerable flexibility in the type of
plot specification. A conditioning expression (on the right side of a
| operator) always implies that different panels are used for
each level of the conditioning factor, according to a Trellis
display. If form is a one-sided formula, histograms of the
variable on the right hand side of the formula, before a |
operator, are displayed (the Trellis function histogram is
used). If form is two-sided and both its left and
right hand side variables are numeric, scatter plots are displayed
(the Trellis function xyplot is used). Finally, if form
is two-sided and its left had side variable is a factor, box-plots of
the right hand side variable by the levels of the left hand side
variable are displayed (the Trellis function bwplot is used).
# S3 method for lme
plot(x, form, abline, id, idLabels, idResType, grid, …)
# S3 method for nls
plot(x, form, abline, id, idLabels, idResType, grid, …)an object inheriting from class "lme", representing
a fitted linear mixed-effects model, or from nls, representing
an fitted nonlinear least squares model.
an optional formula specifying the desired type of
plot. Any variable present in the original data frame used to obtain
x can be referenced. In addition, x itself
can be referenced in the formula using the symbol
".". Conditional expressions on the right of a |
operator can be used to define separate panels in a Trellis
display. Default is resid(., type = "p") ~ fitted(.) ,
corresponding to a plot of the standardized residuals versus fitted
values, both evaluated at the innermost level of nesting.
an optional numeric value, or numeric vector of length two. If given as a single value, a horizontal line will be added to the plot at that coordinate; else, if given as a vector, its values are used as the intercept and slope for a line added to the plot. If missing, no lines are added to the plot.
an optional numeric value, or one-sided formula. If given as
a value, it is used as a significance level for a two-sided outlier
test for the standardized, or normalized residuals. Observations with
absolute standardized (normalized) residuals greater than the \(1
- value/2\) quantile of the standard normal distribution are
identified in the plot using idLabels. If given as a one-sided
formula, its right hand side must evaluate to a logical, integer, or
character vector which is used to identify observations in the
plot. If missing, no observations are identified.
an optional vector, or one-sided formula. If given as a
vector, it is converted to character and used to label the
observations identified according to id. If given as a
one-sided formula, its right hand side must evaluate to a vector
which is converted to character and used to label the identified
observations. Default is the innermost grouping factor.
an optional character string specifying the type of
residuals to be used in identifying outliers, when id is a
numeric value. If "pearson", the standardized residuals (raw
residuals divided by the corresponding standard errors) are used;
else, if "normalized", the normalized residuals (standardized
residuals pre-multiplied by the inverse square-root factor of the
estimated error correlation matrix) are used. Partial matching of
arguments is used, so only the first character needs to be
provided. Defaults to "pearson".
an optional logical value indicating whether a grid should
be added to plot. Default depends on the type of Trellis plot used:
if xyplot defaults to TRUE, else defaults to
FALSE.
optional arguments passed to the Trellis plot function.
a diagnostic Trellis plot.
# NOT RUN {
fm1 <- lme(distance ~ age, Orthodont, random = ~ age | Subject)
# standardized residuals versus fitted values by gender
plot(fm1, resid(., type = "p") ~ fitted(.) | Sex, abline = 0)
# box-plots of residuals by Subject
plot(fm1, Subject ~ resid(.))
# observed versus fitted values by Subject
plot(fm1, distance ~ fitted(.) | Subject, abline = c(0,1))
# }
Run the code above in your browser using DataLab