summary.BayesMfp: Calculate and print the summary of a BayesMfp object

In bfp: Bayesian Fractional Polynomials

Calculate and print the summary of a BayesMfp object

Description

Calculate and print the summary of a BayesMfp object, using S3 methods for the class.

Usage

## S3 method for class 'BayesMfp'
summary(object, level=0.95, table=TRUE,
                           shrinkage=NULL, ...)
## S3 method for class 'summary.BayesMfp'
print(x, ...)

Arguments

object	a valid BayesMfp object
x	a return value of summary.BayesMfp
level	credible level for coefficients HPD intervals (default: 0.95)
table	should a data.frame of the models be included? (default)
shrinkage	shrinkage factor used, where NULL defaults to the posterior expected shrinkage factor
...	only used by summary.BayesMfp to pass arguments to as.data.frame.BayesMfp

Value

summary.BayesMfp returns a list with S3 class summary.BayesMfp, where the arguments “call”, “numVisited”, “termNames”, “shiftScaleMax”, “inclusionProbs”, “chainlength” (only for model sampling results) are copied from the attributes of the BayesMfp object, please see its help page for details.

The other elements are:

dataframe	the model overview as data.frame (only if table=TRUE was specified)
localInclusionProbs	local variable inclusion probability estimates
nModels	number of models contained in object

If there are multiple models in object, the list element postProbs contains the exact (for exhaustively explored model spaces) or estimated (if model sampling has been done) posterior model probabilities.

If object contains only one FP model, then this one is summarized in more detail:

level	used credible level for coefficients HPD intervals
shrinkage	used shrinkage factor
summaryMat	matrix with posterior summaries of the single coefficients: “mode” gives the posterior mode, “HPDlower” and “HPDupper” give the boundaries of the HPD intervals with specified credible level
sigma2Sum	posterior summary for the regression variance: again mode, and lower and upper HPD bounds are given in a rowvector.

Note

Note that if you extract the summary of a single model with these functions, you ignore the uncertainty about the shrinkage factor t=g/(g+1) by plugging in the number shrinkage. If you want to incorporate this uncertainty, you must run BmaSamples on this model and call the corresponding method summary.BmaSamples.

Author(s)

Daniel Saban\'es Bov\'e

See Also

summary.BmaSamples

Examples

## generate a BayesMfp object
set.seed(19)

x1 <- rnorm(n=15)
x2 <- rbinom(n=15, size=20, prob=0.5) 
x3 <- rexp(n=15)

y <- rt(n=15, df=2)

test <- BayesMfp(y ~ bfp (x2, max = 4) + uc (x1 + x3), nModels = 100,
                 method="exhaustive")

## summary of multiple models:
summary(test)

## summary of just one model (no. 10):
summary(test[10])

## internal structure is usually not interesting:
str(summary(test[10]))

bfp documentation built on March 19, 2024, 3:07 a.m.