densLPS.object {DALSM} | R Documentation |
densityLPS
An object returned by function densityLPS
: this is a list
with various components related to the estimation of a density with given mean and variance from potentially right- or interval-censored data using Laplace P-splines.
An object returned by densityLPS
has the following elements:
Essential part:
converged
:
logical convergence indicator.
ddist
:
fitted density function.
Hdist
:
fitted cumulative hazard function.
hdist
:
fitted hazard function.
pdist
:
fitted cumulative distribution function.
ymin, ymax
:
assumed values for the support of the distribution.
phi
:
estimated B-spline coefficients for the log-hazard of the error distribution.
U.phi
:
score of the Lagrangian G(\phi|\omega
).
tau
, ltau
:
selected penalty parameter and its logarithm.
est
:
vector containing the estimated/selected (\phi,\log\tau
) parameters.
fixed.phi
:
logical indicating whether the spline parameters were given fixed values or estimated from the data.
phi.ref
:
reference values for the spline parameters with respect to which \phi
is compared during penalization.
BWB
:
Hessian for \phi
without the penalty contribution.
Prec
:
Hessian or posterior precision matrix for \phi
.
Fisher
:
Fisher information for \phi
.
bins, ugrid, du
:
bins (of width 'du') and with midpoints 'ugrid' partitioning the support of the density.
h.grid, H.grid, dens.grid
:
hazard, cumulative hazard and density values at the grid midpoints 'ugrid'.
h.bins, H.bins, dens.bins
:
hazard, cumulative hazard and density values at the bin limits 'bins'.
expected
:
expected number of observations within each bin.
Finfty
:
integrated density value over the considered support.
Mean0, Var0
:
when specified, constrained mean and variance values during estimation.
mean.dist, var.dist
:
mean and variance of the fitted density.
method
:
method used for penaly selection: "evidence" (by maximizing the marginal posterior for \tau
) or "Schall" (Schall's method).
ed
:
effective number of (spline) parameters.
iterations
:
total number of iterations necessary for convergence.
elapsed.time
:
time required for convergence.
Additional elements: the content of the Dens1d.object used when densityLPS was called.
Philippe Lambert p.lambert@uliege.be
Lambert, P. (2021). Fast Bayesian inference using Laplace approximations in nonparametric double additive location-scale models with right- and interval-censored data. Computational Statistics and Data Analysis, 161: 107250. <doi:10.1016/j.csda.2021.107250>