drop1.uncert: Single variable deletions from the uncertainty budget.

Description

drop1 calculates revised combined uncertainty for single variable deletions from an object of class 'uncert'.

Usage

# S3 method for uncert
drop1(object, scope, simplify = TRUE, 
	which=c("% Change", "var", "u", "var.change", "u.change"), ...)
# S3 method for uncertMC
drop1(object, scope, simplify = TRUE, 
	which=c("% Change", "var", "u", "var.change", "u.change"), ...)
#Print and plot methods 
# S3 method for drop1.uncert
print(x, ..., digits=2)
# S3 method for drop1.uncert
plot(x, ..., 
	which=c("% Change", "var", "u", "var.change", "u.change"))

Value

If simplify=FALSE, an object of class ‘drop1.uncert’, consisting of a data frame with row names corresponding to row.names(object$budget), columns corresponding to all possible values of which in the order "var", "u", "var.change", "u.change", "% Change", and an attribute expr containing a copy of the expr

value of the 'uncert' object to which drop1.uncert is applied.

If simplify=TRUE, the column of the above data frame corresponding to which

is returned as a vector with names row.names(object$budget).

Arguments

object

An object of class ‘uncert’ or ‘uncertMC’.

scope

character vector, expression or formula containing the list of variables to be dropped. If missing, all variables in object$budget are taken as scope.

simplify

logical. If simplify=TRUE, the return value is simplified to a named vector. If FALSE, all forms available (see which) are returned in a data frame.

which

logical; controls the form of information returned when simplify=TRUE. Possible values are:

"var": The modified values of $u(y)^2$.

"u"

The modified values of $u(y)$.

"var.change"

The signed changes in $u(y)^2$.

"u.change"

The signed changes in $u(y)$.

"% Change"

The percentage change in $u(y)$.

An object of class ‘drop1.uncert’ returned by drop1.uncert.

...

Further objects passed to other functions.

digits

number of digits used to format the output. See the digits argument of format.

Author

S. L. R. Ellison, s.ellison@lgcgroup.com

Details

By analogy with drop1, drop1.uncert perfoms single variable deletions from the uncertainty budget in object, calculates the resulting uncertainty and returns the results in the form requested by simplify and which.

‘Single variable deletion’ of a variable $x_i$ is equivalent to setting the uncertainty $u(x_i)$ to zero. Note that this also sets covariance terms involving $x_i$ to zero. drop1.uncert does not support the deletion of single terms such as $cov(i, j)$.

In the case of ‘uncertMC’ objects, drop1 currently requires object$MC$x to be present (i.e. uncertMC called with keep.x=TRUE). The uncertMC method does not support correlation.

For which="var.change", which="u.change" and which="% Change" the change on dropping a variable is negative if the uncertainty reduces on removing the variable.

The print method simply prints the output with a header formed from the expr attribute and with '%' appended to the "% Change" column.

The plot method produces a barplot of the chosen data column. A plot for each value in which is produced. Arguments in ‘...’ are passed to barplot. If not already present in ‘...’ a default main title and ylab are used. The expr attribute is shown as marginal text if not NA.

References

None.

Examples

Run this code

  # \dontshow{
  #An example with negative correlation
  x <- list(a=1, b=3, c=2, d=11)
  u <- lapply(x, function(x) x/10)
  u.cor<-diag(1,4)
  u.cor[3,4]<-u.cor[4,3]<- -0.5
  u.form.c<-uncert(~a+b*2+c*3+d/2, x, u, method="NUM", cor=u.cor)
  # }

  #Continuing the example from plot.uncert:
  require(graphics)
  
  d1<-drop1(u.form.c, simplify=FALSE)
  d1
  
  plot(d1)
  
  drop1(u.form.c)         #% change only

Run the code above in your browser using DataLab