breakage (version 1.1-1)

fit.sensitivity.plot: Visualise fit sensitivity for a breakage-resistance model

Description

Plots a contour map of the sensitivity of the residual error of a breakage resistance model to variation of the fitted parameters. This should give some idea of the goodness of the fit.

Usage

fit.sensitivity.plot(data, fit, rho = 51, l = 1000, r.range = 0.015, theta.range = pi/360, steps = 100, nlevels = 200, r.squared = TRUE, bound.at = 0.99, ...)

Arguments

data

Measured breakage data as an XY list or data frame, with data$x holding the breakage distances in microns, and data$y the corresponding resistance in megohms. For compatibility with breakage.plot and break.clust, these columns may instead by named Z and Mohm respectively.

fit

The fitting results list returned by fit.breakage

rho

Resistivity of the filler solution, in ohm centimetres. Must be the same value used for fitting or the plot will be incorrect.

l

The initial length of the pipette tip, in microns. Must be the same value used for fitting or the plot will be incorrect.

r.range

The range above and below the fitted inner radius result, r, to test and plot, in microns.

theta.range

The range above and below the fitted half-cone angle result, theta, to test and plot, in radians.

steps

The total number of steps to evaluate in both directions.

nlevels

The number of contour lines to draw.

r.squared

Whether to plot the error value directly, or transform it to plot the coefficient of determination (or R-squared) instead. The latter is plotted by default. It contains the same information in marginally more convenient form, but its use might be considered misleading if taken as implying a linear model.

bound.at

A level at which to plot a significance contour. This is on top of the nlevels contours automatically positioned by the main plot, and plotted in a different colour.

...

Any additional parameters to be passed to the main contour plot.

Value

x

The x (theta) values of the plotted surface.

y

The y (radius) values of the plotted surface.

z

The z (error) values of the plotted surface.

See Also

fit.breakage,

Examples

# fake up some breakage data
brks <- sort(abs(0.5 + rnorm(n=15, sd=0.5) * 1:15))
res <- resist.breakage(brks, theta=3*pi/180, r=0.04, rho=64) + rnorm(15)
dat <- list(x=brks, y=res)

# fit it
fit <- fit.breakage(dat, do.plot=FALSE)

# plot the sensitivity surface
fit.sensitivity.plot(dat, fit)