Six plots (selected by which) are available: a plot of residual vs
fitted values, the Q-Qplot of normality, a Scale-Location plot of
sqrt(|residuals|) against fitted values. A plot of Cook's distances
versus row labels, a plot of residuals against leverages, and the optimal
effective rank of "OCV", "GCV", "AIC" or "BIC"
method (only if one of these four methods have been chosen in function dblm).
By default, only the first three and 5 are provided.
# S3 method for dblm
plot(x,which=c(1:3, 5),id.n=3,main="",
cook.levels = c(0.5, 1),cex.id = 0.75,
type.pred=c("link","response"),...)
an object of class dblm or dbglm.
if a subset of the plots is required, specify a subset of the numbers 1:6.
number of points to be labelled in each plot, starting with the most extreme.
an overall title for the plot. Only if one of the six plots is selected.
levels of Cook's distance at which to draw contours.
magnification of point labels.
the type of prediction (required only for a dbglm class object).
Like predict.dbglm, the default "link" is on the scale
of the linear predictors; the alternative "response" is on the scale
of the response variable.
other parameters to be passed through to plotting functions.
Boj, Eva <evaboj@ub.edu>, Caballe, Adria <adria.caballe@upc.edu>, Delicado, Pedro <pedro.delicado@upc.edu> and Fortiana, Josep <fortiana@ub.edu>
The five first plots are very useful to the residual analysis and are
the same that plot.lm. A plot of residuals against fitted
values sees if the variance is constant. The qq-plot checks if the residuals
are normal (see qqnorm).
The plot between "Scale-Location" and the fitted values takes the
square root of the absolute residuals in order to diminish skewness.
The Cook's distance against the row labels, measures the effect of deleting a
given observation (estimate of the influence of a data point). Points with a
large Cook's distance are considered to merit closer examination in the analysis.
Finally, the Residual-Leverage plot also shows the most influence points
(labelled by Cook's distance). See cooks.distance.
The last plot, allows to view the "OCV" (just for dblm), "GCV", "AIC"
or "BIC" criterion according to the used rank in the
dblm or dbglm functions, and chosen the minimum. Applies only if
the parameter full.search its TRUE.
Boj E, Delicado P, Fortiana J (2010). Distance-based local linear regression for functional predictors. Computational Statistics and Data Analysis 54, 429-437.
Boj E, Grane A, Fortiana J, Claramunt MM (2007). Selection of predictors in distance-based regression. Communications in Statistics B - Simulation and Computation 36, 87-98.
Cuadras CM, Arenas C, Fortiana J (1996). Some computational aspects of a distance-based model for prediction. Communications in Statistics B - Simulation and Computation 25, 593-609.
Cuadras C, Arenas C (1990). A distance-based regression model for prediction with mixed data. Communications in Statistics A - Theory and Methods 19, 2261-2279.
Cuadras CM (1989). Distance analysis in discrimination and classification using both continuous and categorical variables. In: Y. Dodge (ed.), Statistical Data Analysis and Inference. Amsterdam, The Netherlands: North-Holland Publishing Co., pp. 459-473.
Belsley, D. A., Kuh, E. and Welsch, R. E. (1980). Regression Diagnostics. New York: Wiley.
dblm for distance-based linear models.
dbglm for distance-based generalized linear models.
n <- 64
p <- 4
k <- 3
Z <- matrix(rnorm(n*p),nrow=n)
b <- matrix(runif(p)*k,nrow=p)
s <- 1
e <- rnorm(n)*s
y <- Z%*%b + e
dblm1 <- dblm(y~Z,metric="gower",method="GCV", full.search=FALSE)
plot(dblm1)
plot(dblm1,which=4)
Run the code above in your browser using DataLab