regPlot: regPlot Analysis

Description

Following a call to the lessR function Regression, in which the returned values of the function are saved into an object, allows the default plots generated by Regression to be accessed one at a time. The specific motivation for this function is to allow custom placement of the graphs from the regression analysis from within knitr. Usually the graphics=FALSE parameter is set on the call to Regression within knitr to suppress the normal graphic output that leads to the generation of the graphs at the beginning of the knitr output.

Usage

regPlot(out, type, digits.d=NULL, pred.intervals=TRUE,
         res.sort=c("cooks","rstudent","dffits","off"),
         res.rows=NULL, cooks.cut=1, scatter.coef=NULL, 
         pdf=FALSE, width=5, height=5, manage.gr=FALSE, …)

Arguments

out

The object returned by the lessR function Regression.

type

Type of plot: 1 plots the scatter plot for a single predictor variable, or the scatter plot matrix for multiple predictors. If a single scatter plot, then the confidence and prediction intervals are included. 2 plots the density and histogram of residuals and 3 plots a scatter plot of the residuals with the fitted values.

digits.d

For the Basic Analysis, the number of decimal digits, set by default to at least 3 or the largest number of digits in the values of the response variable plus 1.

pred.intervals

If set to FALSE, the scatter plot for a single predictor with the response does not contain prediction and confidence intervals.

res.sort

Default is "cooks", for specifying Cook's distance as the sort criterion for the display of the rows of data and associated residuals. Other values are "rstudent" for externally Studentized residuals, "dffits" for dffits and "off" to not sort the rows of data.

res.rows

Default is 20, which lists the first 20 rows of data sorted by the specified sort criterion. To disable residuals, specify a value of 0. To see the output for all observations, specify a value of "all".

cooks.cut

Cutoff value of Cook's Distance at which observations with a larger value are flagged in red and labeled in the resulting scatterplot of Residuals and Fitted Values. Default value is 1.0.

scatter.coef

Display the correlation coefficients in the upper triangle of the scatterplot matrix.

pdf

If TRUE, then graphics are written to pdf files.

width

Width of the pdf file in inches.

height

Height of the pdf file in inches.

manage.gr

Usually leave FALSE. Refers to graphic management of the lessR system.

…

Other parameter values for R function lm which provides the core computations.

Details

OVERVIEW The ability to separate plots is particularly useful with knitr to break up the output to intersperse comments between the plots. For Plot 1, for single predictor a scatter plot with the regression line and confidence and prediction intervals is produced. Otherwise a scatter plot matrix of all the variables in the models is obtained.

To help assess the validity of the model, Plot 2 is of the distribution of the residuals, histogram and density plots, both general and normal. Plot 3 plots the residuals against the fitted value and also identifies the points with the largest values of Cook's distance.

References

Gerbing, D. W. (2014). R Data Analysis without Programming, Chapters 9 and 10, NY: Routledge.

Examples

Run this code

# NOT RUN {
# read internal data set
mydata <- rd("Reading", in.lessR=TRUE, quiet=TRUE)
# do regression analysis, save result into out
reg.out <- reg(Reading ~ Verbal)
# The full output already contains these plots, obtained by
# entering the name of the saved object
reg.out
# Particularly for knitr it is useful to obtain the plots
#   separately from the full output
# Get the scatter plot of the data with the regression line
#   and prediction and confidence intervals 
regPlot(reg.out, 1)

# Can use with multiple regression for the scatter plot matrix
r <- reg(Reading ~ Verbal + Absent + Income)
regPlot(r, 1, scatter.coef=TRUE)
# }