statforbiology (version 0.9.9)

plotnls: Plotting diagnostics for an nls object

Description

This function is aimed at providing some types of plots to assess the goodness of fit for the selected model. Three plots (selectable by the argument 'which') are currently available: a plot of residuals against fitted values (which = 1), a Normal Q-Q plot (which = 2) and a plot of predicted against expected (line) and observed (symbols). By default, type = 3 is provided. As for the third graph, we can either plot all the data (type= "all") or the group means (type = "means"; the default)

Usage

plotnls(x, type = c("average", "all"),
                     xlim = NULL, gridsize = 100,
                     which = 3, ...)

Value

No return value, it produces a plot

Arguments

x

an object of class 'nls'

type

it can be either "means" or "all". In the first case, the group means are plotted for the third graph. It is only considered when which = 3

xlim

The limits for the x-axis (x1, x2)

gridsize

For 'which = 3', it sets the resolution of the fitted line

which

The type of graph: can be 1, 2 or 3 (see description). It defaults to 3.

...

additional graphical arguments

Author

Andrea Onofri

Details

It mimicks the behaviour of the function plot.lm()

Examples

library(statforbiology)
degradation <- read.csv("https://www.casaonofri.it/_datasets/degradation.csv")
mod <- nls(Conc ~ A*exp(-k*Time),
               start=list(A=100, k=0.05),
               data=degradation)
plotnls(mod, which = 3)