fitPlot: Fitted model plot for an lm, glm, or nls object.

Description

A generic function for constructing a fitted model plot for an lm, glm, or nls object. Supported objects are linear models from simple linear regression (SLR), indicator variable regression (IVR), one-way ANOVA, or two-way ANOVA models; general linear models that are logistic regressions with a binary response; and non-linear regression with a single numerical response variable, at least one continuous explanatory variable and up to two group-factor explanatory variables.

Usage

fitPlot(object, ...)

## S3 method for class 'lm':
fitPlot(object, ...)

## S3 method for class 'SLR':
fitPlot(object, plot.pts = TRUE, pch = 16, col.pt = "black",
  col.mdl = "red", lwd = 3, lty = 1, interval = c("none", "confidence",
  "prediction", "both"), conf.level = 0.95, lty.ci = 2, lty.pi = 3,
  xlab = object$Enames[1], ylab = object$Rname, main = "", ...)

## S3 method for class 'IVR':
fitPlot(object, plot.pts = TRUE, pch = c(16, 21, 15, 22, 17,
  24, c(3:14)), col = "rich", lty = c(1:6, 1:6), lwd = 3,
  interval = c("none", "confidence", "prediction", "both"),
  conf.level = 0.95, xlab = object$Enames[1], ylab = object$Rname,
  main = "", legend = "topright", ...)

## S3 method for class 'POLY':
fitPlot(object, ...)

## S3 method for class 'ONEWAY':
fitPlot(object, xlab = object$Enames[1],
  ylab = object$Rname, main = "", type = "b", pch = 16, lty = 1,
  col = "black", interval = TRUE, conf.level = 0.95,
  ci.fun = iCIfp(conf.level), col.ci = col, lty.ci = 1, ...)

## S3 method for class 'TWOWAY':
fitPlot(object, which, change.order = FALSE,
  xlab = object$Enames[ord[1]], ylab = object$Rname, main = "",
  type = "b", pch = c(16, 21, 15, 22, 17, 24, c(3:14)), lty = c(1:6, 1:6,
  1:6), col = "default", interval = TRUE, conf.level = 0.95,
  ci.fun = iCIfp(conf.level), lty.ci = 1, legend = "topright",
  cex.leg = 1, box.lty.leg = 0, ...)

## S3 method for class 'nls':
fitPlot(object, d, pch = c(19, 1), col.pt = c("black", "red"),
  col.mdl = col.pt, lwd = 2, lty = 1, plot.pts = TRUE,
  jittered = FALSE, ylim = NULL, legend = FALSE,
  legend.lbls = c("Group 1", "Group 2"), ylab = names(mdl$model)[1],
  xlab = names(mdl$model)[xpos], main = "", ...)

## S3 method for class 'glm':
fitPlot(object, ...)

## S3 method for class 'logreg':
fitPlot(object, xlab = names(object$model)[2],
  ylab = names(object$model)[1], main = "", plot.pts = TRUE,
  col.pt = "black", transparency = NULL, plot.p = TRUE, breaks = 25,
  p.col = "blue", p.pch = 3, p.cex = 1, yaxis1.ticks = seq(0, 1, 0.1),
  yaxis1.lbls = c(0, 0.5, 1), yaxis2.show = TRUE, col.mdl = "red",
  lwd = 2, lty = 1, mdl.vals = 50, xlim = range(x), ...)

Arguments

object

An lm or nls object (i.e., returned from fitting a model with either lm or nls).

plot.pts

A logical that indicates (TRUE (default)) whether the points are plotted along with the fitted lines. Set to FALSE to plot just the fitted lines.

pch

A numeric or vector of numerics that indicates what plotting characther codes should be used. In SLR this is the single value to be used for all points. In IVR a vector is used to identify the characters for the levels of the second factor.

col.pt

A string used to indicate the color of the plotted points. Used only for SLR and logistic regression objects.

col.mdl

A string used to indicate the color of the fitted line. Used only for SLR and logistic regression objects.

lwd

A numeric used to indicate the line width of the fitted line.

lty

A numeric or vector of numerics used to indicate the type of line used for the fitted line. In SLR this is a single value to be used for the fitted line. In IVR a vector is used to identify the line types for the levels of the second factor. See

interval

In SLR or IVR, a string that indicates whether to plot confidence (="confidence") or prediction (="prediction") intervals. For a SLR object both can be plotted by using ="both". In one-way or two-way ANOVA, a logic

conf.level

A decimal numeric that indicates the level of confidence to use for confidence and prediction intervals.

lty.ci

a numeric used to indicate the type of line used for the confidence band lines for SLR objects or interval lines for one-way and two-way ANOVA. For IVR, the confidence band types are controlled by lty.

lty.pi

a numeric used to indicate the type of line used for the prediction band lines for SLR objects. For IVR, the prediction band types are controlled by lty. See par.

xlab

a string for labelling the x-axis.

ylab

a string for labelling the y-axis.

main

a string for the main label to the plot. Defaults to the model call.

col

A vector of color names or numbers or the name of a paletted (see details) that indicates what color of points and lines to use for the levels of the first factor in an IVR or the second factor in a two-way ANOVA.

legend

Controls use and placement of the legend. See details.

type

The type of graphic to construct in a one-way and two-way ANOVA. If "b" then points are plotted and lines are used to connect points (DEFAULT). If "p" then only points are used and if "l" then only lines are drawn.

ci.fun

A function used to put error bars on the one-way or two-way ANOVA graphs. The default is to use the internal iCIfp function which will place t-distribution based confidence intervals on the graph. The user can provide alternative functions

col.ci

A vector of color names or numbers or the name of a palette (see details) that indicates what colors to use for the confidence interval bars in one-way and two-way ANOVAs.

which

A character string listing the factor in the two-way ANOVA for which the means should be calculated and plotted. This argument is used to indicate for which factor a main effects plot should be constructed. If left missing then an interaction plot is co

change.order

A logical that is used to change the order of the factors in the lm object. This is used to change which factor is plotted on the x-axis and which is used to connect the means when constructing an interaction plot (ignored if which

cex.leg

A single numeric values used to represent the character expansion value for the legend. Ignored if legend=FALSE.

box.lty.leg

A single numeric values used to indicate the type of line to use for the box around the legend. The default is to not plot a box.

A data frame that contains the variabls used in construction of the nls object.

jittered

A logical that indicates whether the points should be jittered horizontally.

ylim

A vector of length two to control the y-axis in the nonlinear regression plot.

legend.lbls

A vector of strings that will be the labels for the legend in an nls fitPlot graphic.

transparency

A numeric that indicates how many points would be plotted on top of each other in a logistic regression before the point would have the full pt.col color. The reciprocal of this value is the alpha transparency value.

plot.p

A logical that indicates if the proportion for categorized values of X are plotted (TRUE; default).

breaks

A number that indicates how many intervals over which to compute proportions or a numeric vector that contains the endpoints of the intervals over which to compute proportions if plot.p=TRUE.

p.col

A color to plot the proportions.

p.pch

A plotting character for plotting the proportions.

p.cex

A character expansion factor for plotting the proportions.

yaxis1.ticks

A numeric vector that indicates where tick marks should be placed on the left y-axis (for the proportion of successes) for the logistic regression plot.

yaxis1.lbls

A numeric vector that indicates labels for the tick marks on the left y-axis (for the proportion of successes) for the logistic regression plot.

yaxis2.show

A logical that indicates whether the right y-axis should be created (=TRUE; default) or not for the logistic regression plot.

mdl.vals

A numer representing the number of values to use for plotting the logistic regression. A larger number means a smoother line.

xlim

A vector of length two to control the x-axis in the logistic regression plot. If this is changed from the default then the domain over which the logistic regression model is plotted will change.

...

Other arguments to be passed to the plot functions.

Value

None. However, a fitted-line plot is produced.

Details

This function does not work with a multiple linear regression, indicator variable regressions with more than two factors, ANOVAs other than one-way and two-way, or models with a categorical response variable. In addition, if the linear model contains a factor then the model must be fit with the quantitative explanatory variable first, followed by the factor(s). This function only works for non-linear models with two or fewer groups. This function is basically a wrapper to a variety of other functions. For one-way or two-way ANOVAs the primary functions called are interaction.plot and lineplot.CI. For simple linear regression the function performs similarly to abline except that the line is constrained to the domain. For indicator variable regression the function behaves as if several abline functions had been called. A legend can be added to the plot in three different ways. First, if legend = TRUE then the R console is suspended until the user places the legend on the graphic by clicking on the graphic at the point where the upper-left corner of the legend should appear. Second, the legend= argument can be set to one of "bottomright", "bottom", "bottomleft", "left", "topleft", "top", "topright", "right" and "center". In this case, the legend will be placed inside the plot frame at the given location. Finally, the legend= argument can be set to a vector of length two which identifies the plot coordinates for the upper-left corner of where the legend should be placed. A legend will not be drawn if legend = FALSE or legend = NULL. A legend also will not be drawn if there are not multiple groups in the model.

Examples

Run this code

## Linear models example
data(Mirex)
Mirex$year <- factor(Mirex$year)

# Indicator variable regression with two factors
lm1 <- lm(mirex~weight*year*species,data=Mirex)
fitPlot(lm1)
fitPlot(lm1,ylim=c(0,0.65),legend="topleft")

# Indicator variable regression with one factor (also showing confidence bands)
lm2 <- lm(mirex~weight*year,data=Mirex)
fitPlot(lm2,legend="topleft")
fitPlot(lm2,legend="topleft",interval="c")

# Indicator variable regression with one factor (assuming parallel lines)
lm3 <- lm(mirex~weight+year,data=Mirex)
fitPlot(lm3,legend="topleft")
fitPlot(lm3,legend="topleft",pch=5)

# Simple linear regression (showing color change and confidence and prediction bands)
lm4 <- lm(mirex~weight,data=Mirex)
fitPlot(lm4,col.mdl="blue")
fitPlot(lm4,interval="both")

# One-way ANOVA
lm5 <- lm(mirex~year,data=Mirex)
fitPlot(lm5)

# Two-way ANOVA (showing both interaction plots and a color change)
lm6 <- lm(mirex~year*species,data=Mirex)
fitPlot(lm6,legend="bottomleft")
fitPlot(lm6,change.order=TRUE)
fitPlot(lm6,change.order=TRUE,col="jet")

# Two-way ANOVA (showing main effects plots)
fitPlot(lm6,which="species")
fitPlot(lm6,which="year")

# Polynomial regression
lm7 <- lm(mirex~weight+I(weight^2),data=Mirex)
fitPlot(lm7,interval="both")

## Non-linear model example
data(Ecoli)
lr.sv <- list(B1=6,B2=7.2,B3=-1.5)
nl1 <- nls(cells~B1/(1+exp(B2+B3*days)),start=lr.sv,data=Ecoli)
fitPlot(nl1,Ecoli,cex.main=0.7,lwd=2)
fitPlot(nl1,Ecoli,xlab="Day",ylab="Cellsx10^6/ml",plot.pts=FALSE)


## Logistic regression example
## NASA space shuttle o-ring failures -- from graphics package
fail <- factor(c(2,2,2,2,1,1,1,1,1,1,2,1,2,1,1,1,1,2,1,1,1,1,1),
levels = 1:2, labels = c("no","yes"))
temperature <- c(53,57,58,63,66,67,67,67,68,69,70,70,70,70,72,73,75,75,76,76,78,79,81)
d <- data.frame(fail,temperature)
glm1 <- glm(fail~temperature,data=d,family="binomial")
fitPlot(glm1)
fitPlot(glm1,breaks=seq(52,82,2))
fitPlot(glm1,yaxis1.ticks=c(0,1),yaxis1.lbls=c(0,1))
# changing the size of the y-axis labels
par(cex.axis=1.5,cex.lab=1.5)
fitPlot(glm1)

Run the code above in your browser using DataLab