deltaMethod: Delta method on data frames

In rpruim/deltaMethod: An Extension of car's Delta Method Functionality

Delta method on data frames

Description

An expansion of the capabilities of deltaMethod from the car package.

Usage

## S3 method for class 'data.frame'
deltaMethod(
  object,
  g,
  uncertainties,
  estimates = measurements,
  func = g,
  constants = c(),
  measurements = NULL,
  vcov.,
  ...
)

Arguments

object	a data frame containing measured quantities
g	a quoted string that is describes the function of the parameter estimates to be evaluated; see deltaMethod for details.
uncertainties	a data frame with the same dimension as object or a numeric vector containing the uncertainties on each measured value in object or a matrix providing a variance-covariance matrix for the uncertainties. If a named vector, and estimates is NULL, the names of the vector will be used for estimates. This makes it possible to specify only object, g, and uncertainties to handle many situations. Alternatvely, if estimates is not NULL, then uncertainties may be a vector of names or integers used to select columns from object. There is one potentially ambiguous case: It is not possible to specify the uncertainties as a vector of integers if estimates is not NULL – such integers will be treated as column numbers for subsetting. If uncertaintites is not a matrix, independece is assumed and the variance-covariance matrix is created under that assumption. Matching of uncertainties to measured values is by position, so names are irrelevant. Uncertainties will be converted into a covariance matrix assuming independence.
estimates	a vector of column names or column numbers used to specify a subset of object containing the measured/estimated quantities.
func	a quoted string used to annotate output. The default of func = g is usually appropriate.
constants	This argument is a named vector whose elements are constants that are used in the f argument. This is needed only when the function is called from within another function to comply to R scoping rules.
measurements	an alternative name for estimates
vcov.	a covariance matrix or a list of covariance matrices. Only one of vcov. and uncertainties may be defined.
...	additional arguments passed through to deltaMethod in the car package.

See Also

deltaMethod in the car package.

Examples

if (require(mosaicData)) {
  C_p <- 4.182 / 60 # / 60 because measureing m in L/min
  exprforQ <- "(T.cold.out - T.cold.in) * C_p * m.cold"
  deltaMethod( HeatX[, c("T.cold.in","T.cold.out","m.cold")], exprforQ, c(1,1,.5) )
  # This is just wordier in this example, but would allow the uncertainties to vary
  # from row to row.

  HeatX3 <-
    transform(HeatX,
      u.cold.in=1, u.cold.out=1, u.hot.in=1, u.hot.out=1,
      u.m.cold=0.5, u.m.hot=0.5)
  deltaMethod( HeatX3[, c("T.cold.in","T.cold.out","m.cold")], exprforQ,
               HeatX3[, c("u.cold.in", "u.cold.out", "u.m.cold")])
  # Rather than specifying two data frames, we can use subsetting instead
  deltaMethod( HeatX3,  exprforQ,
     estimates=c("T.cold.in","T.cold.out","m.cold"),
     uncertainties=c("u.cold.in", "u.cold.out", "u.m.cold"))
  # Can also specify vcov. as a matrix or list of matrices:
  deltaMethod(HeatX[, c("T.cold.in","T.cold.out","m.cold")], exprforQ,
    vcov. = diag(c(1,1,.5)^2) )
  deltaMethod(HeatX[, c("T.cold.in","T.cold.out","m.cold")], exprforQ,
    vcov. = list( diag(c(1,1,.5)^2), diag(c(1,2,.8)^2) ) )
}

rpruim/deltaMethod documentation built on Jan. 21, 2023, 7:59 p.m.