This vignette demonstrates how to
access most of data stored in a stanfit object. A stanfit object (an
object of class "stanfit"
) contains the output derived from
fitting a Stan model using Markov chain Monte Carlo or one of Stan’s
variational approximations (meanfield or full-rank). Throughout the
document we’ll use the stanfit object obtained from fitting the Eight
Schools example model:
[1] "stanfit"
attr(,"package")
[1] "rstan"
There are several functions that can be used to access the draws from
the posterior distribution stored in a stanfit object. These are
extract
, as.matrix
,
as.data.frame
, and as.array
, each of which
returns the draws in a different format.
The extract
function (with its default arguments)
returns a list with named components corresponding to the model
parameters.
[1] "mu" "tau" "eta" "theta" "lp__"
In this model the parameters mu
and tau
are
scalars and theta
is a vector with eight elements. This
means that the draws for mu
and tau
will be
vectors (with length equal to the number of post-warmup iterations times
the number of chains) and the draws for theta
will be a
matrix, with each column corresponding to one of the eight
components:
[1] 1.636743 10.106585 3.513336 11.023126 8.338745 17.326182
[1] 9.572745 1.990020 5.071862 5.896383 21.024283 2.244242
iterations [,1] [,2] [,3] [,4] [,5] [,6]
[1,] 1.8772295 11.129563 1.580120 -4.820752 2.910281 -3.193219
[2,] 11.1700082 10.736322 7.949672 9.920369 7.585557 10.378372
[3,] 0.5476151 6.849613 2.806382 3.297504 12.099958 8.499055
[4,] 15.9757297 16.946180 16.896805 18.205883 5.337248 10.420042
[5,] 28.6057365 12.585173 -5.922589 -1.343499 11.377180 22.199806
[6,] 21.3575933 15.047661 16.108431 16.484844 13.623674 14.455254
iterations [,7] [,8]
[1,] 16.759992 9.074250
[2,] 9.831815 13.308429
[3,] 5.489958 -6.017911
[4,] 13.393023 18.655358
[5,] 6.437247 3.896331
[6,] 17.731023 15.110924
The as.matrix
, as.data.frame
, and
as.array
functions can also be used to retrieve the
posterior draws from a stanfit object:
[1] "mu" "tau" "eta[1]" "eta[2]" "eta[3]" "eta[4]"
[7] "eta[5]" "eta[6]" "eta[7]" "eta[8]" "theta[1]" "theta[2]"
[13] "theta[3]" "theta[4]" "theta[5]" "theta[6]" "theta[7]" "theta[8]"
[19] "lp__"
[1] "mu" "tau" "eta[1]" "eta[2]" "eta[3]" "eta[4]"
[7] "eta[5]" "eta[6]" "eta[7]" "eta[8]" "theta[1]" "theta[2]"
[13] "theta[3]" "theta[4]" "theta[5]" "theta[6]" "theta[7]" "theta[8]"
[19] "lp__"
$iterations
NULL
$chains
[1] "chain:1" "chain:2" "chain:3" "chain:4"
$parameters
[1] "mu" "tau" "eta[1]" "eta[2]" "eta[3]" "eta[4]"
[7] "eta[5]" "eta[6]" "eta[7]" "eta[8]" "theta[1]" "theta[2]"
[13] "theta[3]" "theta[4]" "theta[5]" "theta[6]" "theta[7]" "theta[8]"
[19] "lp__"
The as.matrix
and as.data.frame
methods
essentially return the same thing except in matrix and data frame form,
respectively. The as.array
method returns the draws from
each chain separately and so has an additional dimension:
[1] 4000 19
[1] 4000 19
[1] 1000 4 19
By default all of the functions for retrieving the posterior draws
return the draws for all parameters (and generated quantities).
The optional argument pars
(a character vector) can be used
if only a subset of the parameters is desired, for example:
parameters
iterations mu theta[1]
[1,] 7.935440 9.625193
[2,] 6.636731 7.714632
[3,] 2.077142 5.929779
[4,] 13.197995 11.888387
[5,] 11.083340 6.367072
[6,] 10.186964 10.764070
Summary statistics are obtained using the summary
function. The object returned is a list with two components:
[1] "summary" "c_summary"
In fit_summary$summary
all chains are merged whereas
fit_summary$c_summary
contains summaries for each chain
individually. Typically we want the summary for all chains merged, which
is what we’ll focus on here.
The summary is a matrix with rows corresponding to parameters and
columns to the various summary quantities. These include the posterior
mean, the posterior standard deviation, and various quantiles computed
from the draws. The probs
argument can be used to specify
which quantiles to compute and pars
can be used to specify
a subset of parameters to include in the summary.
For models fit using MCMC, also included in the summary are the Monte
Carlo standard error (se_mean
), the effective sample size
(n_eff
), and the R-hat statistic (Rhat
).
mean se_mean sd 2.5% 25%
mu 8.080801408 0.11205740 5.2075056 -2.0749510 4.5926859
tau 6.468872365 0.14314978 5.5361712 0.2182851 2.4211382
eta[1] 0.364483616 0.01589541 0.9397065 -1.5142246 -0.2529054
eta[2] 0.002946661 0.01451296 0.8814964 -1.7588001 -0.5833192
eta[3] -0.198930820 0.01382374 0.9278652 -2.0142361 -0.8138516
eta[4] -0.048173578 0.01397762 0.8957642 -1.8170455 -0.6409908
eta[5] -0.365129872 0.01415966 0.8797499 -2.0837888 -0.9449596
eta[6] -0.212589063 0.01578280 0.9153735 -1.9886084 -0.8376942
eta[7] 0.328016836 0.01455812 0.9175026 -1.5678663 -0.2498630
eta[8] 0.046709776 0.01409150 0.9253261 -1.8179771 -0.5679017
theta[1] 11.296863420 0.15946130 8.2286601 -1.8944536 5.9899087
theta[2] 7.936680090 0.09610888 6.4394188 -4.8882366 3.8449608
theta[3] 6.132254370 0.13245875 7.8714581 -11.9830147 2.0490261
theta[4] 7.695826305 0.10425387 6.6798046 -5.6843890 3.5682426
theta[5] 5.137745151 0.10251973 6.4321704 -9.2304525 1.4408935
theta[6] 6.392899799 0.10370934 6.7181595 -8.4261449 2.5326189
theta[7] 10.658276186 0.11513742 6.9407663 -2.0018650 6.1114540
theta[8] 8.444759974 0.13676735 7.6950799 -6.5072939 3.9432244
lp__ -39.664060295 0.07628426 2.7152740 -45.8245041 -41.2916710
50% 75% 97.5% n_eff Rhat
mu 7.999463336 11.4383465 18.609302 2159.625 1.0012826
tau 5.182528278 8.8589141 20.108089 1495.677 1.0004633
eta[1] 0.396314473 0.9967559 2.144416 3494.949 0.9996598
eta[2] -0.002458377 0.5935754 1.717753 3689.173 0.9999393
eta[3] -0.213413961 0.4125764 1.627728 4505.245 0.9996014
eta[4] -0.051268014 0.5497469 1.694867 4106.966 0.9997956
eta[5] -0.366101365 0.1993965 1.384328 3860.228 1.0003907
eta[6] -0.212427122 0.4001244 1.570076 3363.789 1.0013953
eta[7] 0.342566320 0.9224976 2.114650 3971.950 0.9997084
eta[8] 0.050910504 0.6819199 1.834986 4311.964 0.9996240
theta[1] 10.361269440 15.3604216 31.235137 2662.856 1.0000660
theta[2] 7.966544858 12.0469553 20.726894 4489.173 1.0000285
theta[3] 6.673599321 11.0517115 20.033778 3531.418 0.9995364
theta[4] 7.772971015 11.7123420 21.177812 4105.283 1.0001019
theta[5] 5.569444083 9.4745957 16.695643 3936.409 1.0008361
theta[6] 6.764191552 10.5941905 18.886780 4196.284 0.9997453
theta[7] 10.190632594 14.4894225 26.060456 3633.974 0.9995850
theta[8] 8.252979095 12.8373335 25.323362 3165.639 0.9993903
lp__ -39.359995581 -37.7498956 -34.961036 1266.944 1.0007259
If, for example, we wanted the only quantiles included to be 10% and
90%, and for only the parameters included to be mu
and
tau
, we would specify that like this:
mu_tau_summary <- summary(fit, pars = c("mu", "tau"), probs = c(0.1, 0.9))$summary
print(mu_tau_summary)
mean se_mean sd 10% 90% n_eff Rhat
mu 8.080801 0.1120574 5.207506 1.582639 14.43934 2159.625 1.001283
tau 6.468872 0.1431498 5.536171 1.033303 13.80869 1495.677 1.000463
Since mu_tau_summary
is a matrix we can pull out columns
using their names:
10% 90%
mu 1.582639 14.43934
tau 1.033303 13.80869
For models fit using MCMC the stanfit object will also contain the
values of parameters used for the sampler. The
get_sampler_params
function can be used to access this
information.
The object returned by get_sampler_params
is a list with
one component (a matrix) per chain. Each of the matrices has number of
columns corresponding to the number of sampler parameters and the column
names provide the parameter names. The optional argument inc_warmup
(defaulting to TRUE
) indicates whether to include the
warmup period.
sampler_params <- get_sampler_params(fit, inc_warmup = FALSE)
sampler_params_chain1 <- sampler_params[[1]]
colnames(sampler_params_chain1)
[1] "accept_stat__" "stepsize__" "treedepth__" "n_leapfrog__"
[5] "divergent__" "energy__"
To do things like calculate the average value of
accept_stat__
for each chain (or the maximum value of
treedepth__
for each chain if using the NUTS algorithm,
etc.) the sapply
function is useful as it will apply the
same function to each component of sampler_params
:
mean_accept_stat_by_chain <- sapply(sampler_params, function(x) mean(x[, "accept_stat__"]))
print(mean_accept_stat_by_chain)
[1] 0.8871174 0.9019111 0.8683277 0.8614057
max_treedepth_by_chain <- sapply(sampler_params, function(x) max(x[, "treedepth__"]))
print(max_treedepth_by_chain)
[1] 4 5 4 4
The Stan program itself is also stored in the stanfit object and can
be accessed using get_stancode
:
The object code
is a single string and is not very
intelligible when printed:
[1] "data {\n int<lower=0> J; // number of schools\n real y[J]; // estimated treatment effects\n real<lower=0> sigma[J]; // s.e. of effect estimates\n}\nparameters {\n real mu;\n real<lower=0> tau;\n vector[J] eta;\n}\ntransformed parameters {\n vector[J] theta;\n theta = mu + tau * eta;\n}\nmodel {\n target += normal_lpdf(eta | 0, 1);\n target += normal_lpdf(y | theta, sigma);\n}"
attr(,"model_name2")
[1] "schools"
A readable version can be printed using cat
:
data {
int<lower=0> J; // number of schools
real y[J]; // estimated treatment effects
real<lower=0> sigma[J]; // s.e. of effect estimates
}
parameters {
real mu;
real<lower=0> tau;
vector[J] eta;
}
transformed parameters {
vector[J] theta;
theta = mu + tau * eta;
}
model {
target += normal_lpdf(eta | 0, 1);
target += normal_lpdf(y | theta, sigma);
}
The get_inits
function returns initial values as a list
with one component per chain. Each component is itself a (named) list
containing the initial values for each parameter for the corresponding
chain:
$mu
[1] -1.44699
$tau
[1] 0.9667098
$eta
[1] -1.2508485 0.7433085 -0.3798570 -1.0922688 -0.7170482 -0.2388567 -1.5208663
[8] 0.1987448
$theta
[1] -2.6561980 -0.7284269 -1.8142020 -2.5028974 -2.1401680 -1.6778956 -2.9172269
[8] -1.2548620
The get_seed
function returns the (P)RNG seed as an
integer:
[1] 1899244817