gdensity {BMS} | R Documentation |
Calculates the mixture marginal posterior density for the shrinkage factor (g/(1+g)) from a BMA object under the hyper-g prior and plots it
gdensity(x, n = 512, plot = TRUE, addons = "zles", addons.lwd = 1.5, ...)
x |
A bma object (see |
n |
The integer number of equally spaced points at which the density is to be estimated. see 'Details' below |
addons |
character, defaulting to |
plot |
logical. If |
addons.lwd |
scalar, default 1.5. Line width to be used for the low-level plotting commands specified by |
... |
Additional arguments for |
The function gdensity
estimates and plots the posterior density for the shrinkage factor g/(1+g)
This is evidently only possible if the shrinkage factor if not fixed, i.e. if the bma object x
was estimated with a hyper-g prior - cf. argument g
in bms
The density is based only on the best models retained in the bma object x
, cf. argument nmodel
in bms
A note on argument n
: The points at which the density is estimated start at max(0,E-5*SD), where E and SD are the expected value and standard deviation of the shrinkage factor, respectively. For plotting the entire domain (0,1) use xlim=c(0,1)
as an argument for gdensity
.
The argument addons
specifies what additional information should be added to the plot(s) via the low-level commands lines
and legend
:
"e"
for the posterior expected value (EV) of the shrinkage factor,
"s"
for 2 times posterior standard deviation (SD) bounds,
"m"
for the posterior median,
"f"
for posterior expected values of the individual models whom the density is averaged over,
"z"
for a zero line, "l"
for including a legend
The following two are only possible if the bma object collected statistics on shrinkage, cf. argument g.stats
in bms
"E"
for posterior expected value under MCMC frequencies (see argument exact
in coef.bma
),
"S"
for the corresponding 2 times standard deviation bounds (MCMC),
Any combination of these letters will give the desired result. Use addons=""
for not using any of these.
gdensity
returns an object of the class density
detailing the posterior mixture density of the shrinkage factor.
The computed marginal posterior density is a Bayesian Model Averaging mixture of the marginal posterior densities of the shrinkage factor under individual models.
The accuracy of the result therefore depends on the number of 'best' models contained in x
(cf. argument nmodel
in bms
).
Correspondingly, the posterior EV and SD specified by addons="es"
are based on 'best' model likelihoods ('exact') and are conditional on inclusion.
The low-level commands enacted by the argument addons
rely on colors of the palette
: color 2 for "e"
and "s"
, color 3 for "m"
, color 8 for "f"
, color 4 for "E"
and "S"
. The default colors may be changed by a call to palette
.
Martin Feldkircher and Stefan Zeugner
density.bma
for computing coefficient densities, bms
for creating bma objects, density
for the general method
Check http://bms.zeugner.eu for additional help.
data(datafls) mm=bms(datafls,g="hyper=UIP") gdensity(mm) # default plotting # the grey bars represent expected shrinkage factors of the individual models gdensity(mm,addons="lzfes") # #plotting the median 'm' and the posterior mean and bounds based on MCMC results: gdensity(mm,addons="zSEm",addons.lwd=2) # plot the posterior shrinkage density only for the very best model gdensity(mm[1],addons="esz") #using the calculated density for other purposes... dd=gdensity(mm,plot=FALSE) plot(dd)