plotReg: Plot regression

In Pakillo/pacotools: Miscellaneous tools I often need and forget about

Description

Visualise fitted regression and data with uncertainty (either a polygon or spaghetti plot)

Usage

plotReg(xdata, ydata, betamat, predmat, predcol, plot.unc = T, spag = T,
  colour, alpha.spag = 1, lwd.spag = 0.1, alpha.polyg = 80, add = F,
  ...)

Arguments

xdata	The predictor. A numeric vector.
ydata	The response variable. A numeric vector (can be a binary variable).
betamat	A numeric matrix of nsims x nparams dimensions with the posterior values of the model parameters, e.g. as retrieved from model$BUGSoutput$sims.matrix[,c("intercept", "slope")]
predmat	A numeric matrix with ncol=nparams, containing the values of the covariates that should be used for the plot, including the intercept. The order of the covariate columns should be the same as the parameter columns in betamat.
predcol	A number telling which column in predmat represents the covariate to plot.
plot.unc	Logical. Plot uncertainty or just the average fitted response?
spag	Logical. Make spaghetti plot? If False, and plot.unc=T, a shaded polygon is drawn instead
colour	Colour of the regression lines. If missing, it is taken from palette "Set1" in RColorBrewer.
alpha.spag	Alpha multiplier. A number between 0 and 1 (default=1) to adjust the transparency of the uncertainty lines in the spaghetti plot.
lwd.spag	Line width (lwd) of every line in the spaghetti plot. Together with alpha.spag, define the degree of shadeness of the spaghetty plot
alpha.polyg	Alpha parameter (values between 0 and 255) for setting the transparency of the shaded polygon.
add	Logical. Add another regression line to an existing plot?
...	Further parameters to pass to plot. See par

Value

A plot

Author(s)

Paco, based on code by Lucas Leeman (see http://andrewgelman.com/2012/08/graphs-showing-regression-uncertainty-the-code/)

Examples

### LINEAR REGRESSION ###

# First generate some data
interc <- 0.6
slope <- 0.2
sigma <- 0.1
x <- seq(from=1, to=10, length.out=100)
y <- rnorm(100, interc + slope*x, sigma)

xdata <- x
ydata <- y
nsims <- 200
interc.post <- rnorm(nsims, interc, 0.05)
slope.post <- rnorm(nsims, slope, 0.01)
betamat <- cbind(interc.post, slope.post)
predmat <- cbind(rep(1,length(xdata)), xdata)

#plotReg(xdata, ydata, betamat, predmat)
#plotReg(xdata, ydata, betamat, predmat, spag=F)
#plotReg(xdata, ydata, betamat, predmat, plot.unc=F)
#plotReg(xdata, ydata, betamat, predmat, main="My regression", xlab="My x")    # including graphical parameters
#plotReg(xdata, ydata, betamat, predmat, colour="green")    # specifying colour

## Add another regression lines to the plot

### LOGISTIC REGRESSION ###
# Generate some data
x <- seq(1:1000)
yhat <- plogis(0.2 + 0.003*x)   # inverse logit
y <- rbinom(1000, 1, yhat)

# Fit model and get predicted probabilities
model <- glm(y~x, family="binomial")
ypred <- predict(model, type="response")

Pakillo/pacotools documentation built on May 7, 2019, 11:56 p.m.