plot.eff.biv: Plot method for eff.biv objects

In RXshrink: Maximum Likelihood Shrinkage using Generalized Ridge or Least Angle Regression

Plot method for eff.biv objects

Description

Plot One or Two Bivariate Confidence Ellipses or a TRACE display for efficiently shrunken regression coefficients. The default is to display two ellipses at confidence levels 0.95 and 0.50 on a single plot for the first 2 of p >= 2 coefficients. The projection of the eff.ridge() Path from the OLS solution (at m = 0) to the Knot at the most likely Minimum MSE Risk solution (at m = mStar) ...plus the continuation of this Path towards the Shrinkage Terminus at (0,0) is shown in "red". To show the (outer) Ellipse as large as possible, the (0,0) point may be "off the plot".

Usage

  ## S3 method for class 'eff.biv'
plot(x, type = "ellip", ...)

Arguments

x	Output list object of class eff.ridge.
type	One of 2 options: "ellip" or "trace". The default option of "ellip" displays the Confidence Ellipse(s) specified by arguments to eff.biv().
...	Optional argument(s) passed on to plot().

Value

NULL

Author(s)

Bob Obenchain <[e-mail]>

Examples

  # Cost-Effectiveness inferences using Linear Models and eff.ridge...
  ## Not run: 
    library(ICEinfer)
    data(sepsis)
    ndr <- ICEpref(sepsis$icu, sepsis$qalypres, sepsis$totcost, lambda=50000, beta=0.1)
    sndr <- data.frame(cbind(ndr$pref, sepsis)) # ndr: non-linear diminishing returns...
    form4 <- ndr.pref ~ icu + age + orgfails + apache
    usndra <- eff.aug(eff.ridge(form4, sndr)) # compare ndr of 2 Intensive Care Units...
    plot(efobj <- eff.biv(usndra, 2, 4))
    efobj          # implicit print...
    # plot(efobj, type = "tr")
  
## End(Not run)

RXshrink documentation built on Aug. 8, 2023, 1:09 a.m.