version 0.9.10

DESCRIPTION

@@ -1,19 +1,19 @@
 Package: bsts
-Version: 0.9.9
-Date: 2022-11-03
+Version: 0.9.10
+Date: 2024-01-16
 Title: Bayesian Structural Time Series
 Author: Steven L. Scott <[email protected]>
 Maintainer: Steven L. Scott <[email protected]>
 Description: Time series regression using dynamic linear models fit using
  MCMC. See Scott and Varian (2014) <DOI:10.1504/IJMMNO.2014.059942>, among many
  other sources.
-Depends: BoomSpikeSlab (>= 1.2.5), zoo (>= 1.8), xts, Boom (>= 0.9.11),
+Depends: BoomSpikeSlab (>= 1.2.6), zoo (>= 1.8), xts, Boom (>= 0.9.13),
  R(>= 3.4.0)
 Suggests: testthat
-LinkingTo: Boom (>= 0.9.11)
+LinkingTo: Boom (>= 0.9.13)
 License: LGPL-2.1 | MIT + file LICENSE
 Encoding: UTF-8
 NeedsCompilation: yes
-Packaged: 2022-11-06 20:47:53 UTC; steve
+Packaged: 2024-01-16 23:31:15 UTC; steve
 Repository: CRAN
-Date/Publication: 2022-11-07 08:50:11 UTC
+Date/Publication: 2024-01-17 13:02:07 UTC

MD5

@@ -1,4 +1,4 @@
-3e7a94e74d8420bd9659550450c174f2 *DESCRIPTION
+9b8057b8ebc35b974db9119feef02663 *DESCRIPTION
 ccd3022578bbd36b69595f56cd8d83e5 *LICENSE
 1570eae7bf1ed9a241feeacb9404345b *NAMESPACE
 ceb07fe9db975f5c42496dfaa33a5d14 *R/add.ar.R
@@ -53,6 +53,7 @@ abadc45049de045297dc88b776546d08 *inst/tests/tests/testthat/test-poisson.R
 e91ff5f823c6acbff0b7b7d6810a82a8 *inst/tests/tests/testthat/test-prediction.R
 74913baa6a0f8135591ba1c6d7b7e46e *inst/tests/tests/testthat/test-regressionholiday.R
 ebe710afd159f82025b7c8b9d6e6d237 *inst/tests/tests/testthat/test-seasonal.R
+e0c7763b06a704f13b10d4248db8f344 *inst/tests/tests/testthat/test-sigma-upper-limit.R
 6c0c26d250eb86a7f004138b65297c90 *inst/tests/tests/testthat/test-student.R
 9f6f10bf2045923b08f20734cef76b22 *inst/tests/tests/testthat/test-trig.R
 0465bac211a2e128119a6fbd17f19c56 *inst/tests/testthat.R
@@ -70,6 +71,7 @@ abadc45049de045297dc88b776546d08 *inst/tests/testthat/test-poisson.R
 e91ff5f823c6acbff0b7b7d6810a82a8 *inst/tests/testthat/test-prediction.R
 74913baa6a0f8135591ba1c6d7b7e46e *inst/tests/testthat/test-regressionholiday.R
 ebe710afd159f82025b7c8b9d6e6d237 *inst/tests/testthat/test-seasonal.R
+e0c7763b06a704f13b10d4248db8f344 *inst/tests/testthat/test-sigma-upper-limit.R
 6c0c26d250eb86a7f004138b65297c90 *inst/tests/testthat/test-student.R
 9f6f10bf2045923b08f20734cef76b22 *inst/tests/testthat/test-trig.R
 65ad45f30d10d63352989bcab06c428f *man/HarveyCumulator.Rd
@@ -91,7 +93,8 @@ dd0f4a731a23a55e13853e991fd64d2c *man/add.trig.Rd
 98e85cce5007520dfeb7e492ade2722a *man/aggregate.time.series.Rd
 bd91ad411945df7d74b054e3561cc2a6 *man/aggregate.weeks.to.months.Rd
 db4593d8299a0b801fe53df3ecd777da *man/auto.ar.Rd
-46167431326cedf0a047ae9a8dc18862 *man/bsts.Rd
+0467eabb71ff9bfc6ae4d958d09aae86 *man/bsts-package.Rd
+82ca0c557d1452e4d6abc107365e9a84 *man/bsts.Rd
 177c7d07d4182091f2d0b810a1dd0819 *man/bsts.options.Rd
 2ceaadc5491d127d9021ff75b21c8e7e *man/compare.bsts.models.Rd
 99b12efa80a168f273cf27497dd8ba2b *man/date.range.Rd
@@ -111,7 +114,7 @@ e7b9351ff2f8b0c117c1b2377303603d *man/geometric.sequence.Rd
 cf7b7d18e7252f5eccd0341eb26a6d82 *man/last.day.in.month.Rd
 d001026a7f503a1d7e4c3139b193bfe5 *man/match.week.to.month.Rd
 5c321a7f4189941f9d151f71627da8d2 *man/max.window.width.Rd
-576c4c6a098bada394edefb47521ed30 *man/mbsts.Rd
+ddf15410ad1df7752a9ac9d277b0b409 *man/mbsts.Rd
 43b82cce35af62b441d08cda809329a1 *man/mixed.frequency.Rd
 4c9639c7bfb85396081677060adc4541 *man/month.distance.Rd
 129cf408079a8ce4c6fea7e5e1c54515 *man/named.holidays.Rd
@@ -142,25 +145,25 @@ bdeeea57d63366747d7bbace634387f7 *man/turkish.Rd
 b57c602efad59be679c77a916faf619c *man/week.ends.Rd
 d96e18ecf4ab02527bf1b60ce08f106e *man/weekday.names.Rd
 cd4f3fcff595de3ac83d8e85ce89e627 *man/wide.to.long.Rd
-ad3e4c15cd397b85b1c6a45f0803500a *src/Makevars
+6afd68425262b76be593fe3cff3d8515 *src/Makevars
 d99ce4ef93988dc2e1b7279ab92be31e *src/aggregate_time_series.cc
-84059b88fee889c7194f6d541ce3ed54 *src/bsts.cc
+2fe9977b38f7901775a6f0ae9341eb00 *src/bsts.cc
 8c5caf631398cd06050d059691d443e3 *src/bsts_init.cc
 604673026495a954489f5e86602bc48f *src/create_dynamic_intercept_state_model.cpp
 2d60cd8ba33ef34e0238517e07bf7c9c *src/create_dynamic_intercept_state_model.h
-74533455e40f6d52661548ec47182a25 *src/create_shared_state_model.cpp
-06586064da9cccf1136dd127958adefe *src/create_shared_state_model.h
+619fe844a942187bf979040bffe92752 *src/create_shared_state_model.cpp
+2b570f69cfae36e7504ca69b0629812b *src/create_shared_state_model.h
 776cfc2e6f5b39a8354601ceaf73905b *src/create_state_model.cpp
 f8f89c660dc8cf6920e8694d25b99671 *src/create_state_model.h
-4a0ab82799effea8a0f2998cfb9302f7 *src/dirm.cc
+d65049aab4c84ce9c17312853647bbe2 *src/dirm.cc
 c230f6664bd466e38e8e427d73b81cd2 *src/dynamic_intercept_model_manager.cc
 f4b47e0c03ea817b60f652813b16fb06 *src/dynamic_intercept_model_manager.h
 e6f88ebc8a3c65c29fd14268e57fe4d9 *src/get_date_ranges.cc
 cf13c9e432dda0d3468449ec04dc4804 *src/mbsts.cc
 4bc3088809b3d5525ee65288bf8aa959 *src/mixed_frequency.cc
 bc75e43ff4ec384d60484d3adee324ba *src/model_manager.cc
 17d5861e914a1658a6ec202fce330e6a *src/model_manager.h
-c25fa6519d37e732f1ffc9016596a23d *src/multivariate_gaussian_model_manager.cc
+ca7ae7e3f1e12ed16c131187921c2595 *src/multivariate_gaussian_model_manager.cc
 d1cd0b01191824fd3e84d41bcedd276a *src/multivariate_gaussian_model_manager.h
 74e7f2c3bc45e9f519ed32dcfcc4f134 *src/state_space_gaussian_model_manager.cc
 2bedea75559c6e5f0a0bd7d3af6090e8 *src/state_space_gaussian_model_manager.h
@@ -170,27 +173,9 @@ a432aebec2404eb926fe13e9bcb5fb25 *src/state_space_logit_model_manager.cc
 e216be891919e3263441168f35507631 *src/state_space_poisson_model_manager.h
 d12a59708f82f866ed02944f5431bb59 *src/state_space_regression_model_manager.cc
 7ac07cf5d57c124a845c31baeb64f782 *src/state_space_regression_model_manager.h
-0af65cdc6fe068f8fc2df588f4898f05 *src/state_space_student_model_manager.cc
+7f6a474ddee50e7d05fd12687d3dd8fe *src/state_space_student_model_manager.cc
 70d7c3d65ab85979568d692b4a217f9f *src/state_space_student_model_manager.h
 b8f60bed801663863a327aa8bd4d367e *src/timestamp_info.cc
 1c682544f82b2cf9331c827ec6c3f0cf *src/timestamp_info.h
 758355f181bebfed943c020342250e5e *src/utils.cc
 02c97999019ae7fc533919fff4029385 *src/utils.h
-0465bac211a2e128119a6fbd17f19c56 *tests/testthat.R
-2c297be76520a1e86a59e50ffce13d00 *tests/testthat/test-ar.R
-630452660972de113c484c1dd4d52269 *tests/testthat/test-autoar.R
-172df9e68c0f7876dcaade3768c2d5f9 *tests/testthat/test-date-range.R
-0126f04464776f5b1b05c32d25d25f55 *tests/testthat/test-dirm.R
-3de38eb5bcb72ddf38c3b818d9518c0d *tests/testthat/test-dynamic-regression.R
-236c11dceb0ea37b1966a706f7ac3a62 *tests/testthat/test-goog.R
-8e8e34dbc5a1d1735ec663091bcd03a9 *tests/testthat/test-holidays.R
-5c8960cc0a588cb58e49ab39b39cf683 *tests/testthat/test-multivariate.R
-b8cf1c7b59eede2825f50fea5feac4c1 *tests/testthat/test-plot-components.R
-abadc45049de045297dc88b776546d08 *tests/testthat/test-poisson.R
-5c420ef79df05489ab47c1ee48ee0323 *tests/testthat/test-prediction-errors.R
-e91ff5f823c6acbff0b7b7d6810a82a8 *tests/testthat/test-prediction.R
-74913baa6a0f8135591ba1c6d7b7e46e *tests/testthat/test-regressionholiday.R
-ebe710afd159f82025b7c8b9d6e6d237 *tests/testthat/test-seasonal.R
-e0c7763b06a704f13b10d4248db8f344 *tests/testthat/test-sigma-upper-limit.R
-6c0c26d250eb86a7f004138b65297c90 *tests/testthat/test-student.R
-9f6f10bf2045923b08f20734cef76b22 *tests/testthat/test-trig.R

inst/tests/testthat/test-sigma-upper-limit.R

@@ -0,0 +1,19 @@
+library(bsts)
+library(testthat)
+
+test_that("sigma.upper.limit can be specified at the main bsts call.", {
+ y <- rnorm(4)
+ x <- rnorm(4)
+ ss <- AddLocalLevel(list(), y)
+
+ warning("enable the test for 'sigma.upper.limit'")
+
+ ## for (i in 1:20) {
+ ## model <- bsts(y, ss, niter = 100, ping = -1, sigma.upper.limit = 10)
+ ## }
+ ## expect_true(inherits(model, "bsts"))
+
+ ## for(i in 1:20) {
+ ## model <- bsts(y ~ x, ss, niter = 10, ping = -1, sigma.upper.limit = 10)
+ ## }
+})

man/bsts-package.Rd

@@ -0,0 +1,36 @@
+\name{bsts-package}
+\alias{bsts-package}
+\docType{package}
+\title{
+bsts
+}
+
+\description{
+ Time series regression using dynamic linear models fit using
+ MCMC. See Scott and Varian (2014) <DOI:10.1504/IJMMNO.2014.059942>, among many
+ other sources.
+}
+
+\details{
+ \subsection{Installation note for Linux users}{ If you are installing bsts
+ using \code{\link{install.packages}} on a Linux machine (and thus
+ compiling yourself) you will almost certainly want to set the
+ \code{Ncpus} argument to a large number. Windows and Mac users can
+ ignore this advice. }
+
+}
+
+\author{
+Author: Steven L. Scott <steve.the.bayesian@gmail.com>
+Maintainer: Steven L. Scott <steve.the.bayesian@gmail.com>
+}
+
+\references{
+ Please see the references in the help page for the \code{\link{bsts}} function.
+}
+
+\keyword{ package }
+
+\seealso{
+ See the examples in the \code{\link{bsts}} function.
+}

man/bsts.Rd

@@ -211,6 +211,15 @@ bsts(formula,
  exposure times. }
 
 \references{
+ Scott and Varian (2014)
+ "Predicting the Present with Bayesian Structural Time Series",
+ International Journal of Mathematical Modelling and Numerical
+ Optimization. 4--23.
+
+ Scott and Varian (2015)
+ "Bayesian Variable Selection for Nowcasting Economic Time Series",
+ Economic Analysis of the Digital Economy, pp 119-135.
+
  Harvey (1990), "Forecasting, structural time series, and the Kalman
  filter", Cambridge University Press.
 
@@ -220,12 +229,6 @@ bsts(formula,
  George and McCulloch (1997)
  "Approaches for Bayesian variable selection", Statistica Sinica pp
  339--374.
-
- Ghosh and Clyde (2011)
- "Rao-Blackwellization for Bayesian variable selection and model averaging
- in linear and binary regression: a novel data augmentation approach",
- JASA pp 1041 --1052.
-
 }
 
 \author{

man/mbsts.Rd

@@ -273,7 +273,7 @@ opar <- par(mfrow=c(nfactors,1), mar=c(0, 4, 0, 4), omi=rep(.25, 4))
 burn <- 10
 for(j in 1:nfactors) {
  BoxplotTrue(model$shared.local.level.coefficients[-(1:burn), j, ],
- t(observation.coefficients[, j]), axes=F, truth.color="blue")
+ t(observation.coefficients)[, j], axes=F, truth.color="blue")
  abline(h=0, lty=3)
  box()
  axis(2)

src/Makevars

@@ -1,9 +1,15 @@
 BOOM_DIR = `$(R_HOME)/bin/Rscript -e "cat(system.file(package='Boom'))"`
 BOOM_LIB = ${BOOM_DIR}/lib${R_ARCH}/libboom.a
-PKG_LIBS = ${BOOM_LIB} 
+PKG_LIBS = ${BOOM_LIB}
 PKG_CPPFLAGS = -I${BOOM_DIR}/include -DADD_ -DR_NO_REMAP -DRLANGUAGE
 
 # PKG_CXXFLAGS= -fsanitize=address
 # PKG_CXXFLAGS= -fsanitize=undefined
 
-CXX_STD = CXX11
+# CXX_STD = CXX11
+
+# When compiling remotely (e.g. on CRAN's winbuilder) flags to be
+# passed to make can be specified here.
+# -k: keep going
+# -j 16: use 16 threads
+# MAKEFLAGS=" -j 32 "

src/bsts.cc

@@ -98,7 +98,7 @@ extern "C" {
  << "Time used was "
  << double(current_time - start_time) / CLOCKS_PER_SEC
  << " seconds.";
- Rf_warning(warning.str().c_str());
+ Rf_warning("%s", warning.str().c_str());
  return BOOM::appendListElement(
  ans,
  ToRVector(BOOM::Vector(1, i + 1)),

src/create_shared_state_model.cpp

@@ -43,13 +43,15 @@ namespace BOOM {
  SharedStateModelVector &state_models,
  CIMSSMB *model,
  SEXP r_shared_state_specification,
+ const std::vector<Ptr<UnivParams>> &residual_variance_parameters,
  const std::string &prefix) {
  if (!model) return;
  int number_of_state_models = Rf_length(r_shared_state_specification);
  for (int i = 0; i < number_of_state_models; ++i) {
  state_models.add_state(CreateSharedStateModel(
  model,
  VECTOR_ELT(r_shared_state_specification, i),
+ residual_variance_parameters,
  prefix));
  }
  InstallPostStateListElements();
@@ -82,9 +84,14 @@ namespace BOOM {
  CISSMF::CreateSharedStateModel(
  CIMSSMB *model,
  SEXP r_state_component,
+ const std::vector<Ptr<UnivParams>> &residual_variance_parameters,
  const std::string &prefix) {
  if (Rf_inherits(r_state_component, "SharedLocalLevel")) {
- return CreateSharedLocalLevel(r_state_component, model, prefix);
+ return CreateSharedLocalLevel(
+ r_state_component,
+ model,
+ residual_variance_parameters,
+ prefix);
  } else {
  report_error("Unrecognized shared state model.");
  }
@@ -125,6 +132,7 @@ namespace BOOM {
  Ptr<SharedStateModel> CISSMF::CreateSharedLocalLevel(
  SEXP r_state_component,
  CIMSSMB *model,
+ const std::vector<Ptr<UnivParams>> &residual_variance_parameters,
  const std::string &prefix) {
  int nfactors = lround(Rf_asReal(getListElement(r_state_component, "size")));
  NEW(ConditionallyIndependentSharedLocalLevelStateModel, state_model)(
@@ -161,7 +169,8 @@ namespace BOOM {
 
  // Set the posterior sampler for the overall state model.
  NEW(ConditionallyIndependentSharedLocalLevelPosteriorSampler,
- state_model_sampler)(state_model.get(), slabs, spikes);
+ state_model_sampler)(state_model.get(), slabs, spikes,
+ residual_variance_parameters);
 
  state_model->set_method(state_model_sampler);
 

src/create_shared_state_model.h

@@ -80,6 +80,7 @@ namespace BOOM {
  void AddState(SharedStateModelVector &state_models,
  CIMSSMB *model,
  SEXP r_shared_state_specification,
+ const std::vector<Ptr<UnivParams>> &residual_variance_parameters,
  const std::string &prefix = "");
 
  // Save the final state (i.e. at time T) of the model for use with
@@ -118,13 +119,15 @@ namespace BOOM {
  Ptr<SharedStateModel> CreateSharedStateModel(
  CIMSSMB *model,
  SEXP r_state_component,
+ const std::vector<Ptr<UnivParams>> &residual_variance_parameters,
  const std::string &prefix);
 
 
  // Specific functions to create specific state models.
  Ptr<SharedStateModel> CreateSharedLocalLevel(
  SEXP r_state_component,
  CIMSSMB *model,
+ const std::vector<Ptr<UnivParams>> &residual_variance_parameters,
  const std::string &prefix);
  };
 

src/dirm.cc

@@ -35,7 +35,7 @@ extern "C" {
  using namespace BOOM;
  using namespace BOOM::RInterface;
  using namespace BOOM::bsts;
- 
+
  SEXP analysis_common_r_fit_dirm_(
  SEXP r_data_list,
  SEXP r_state_specification,
@@ -55,7 +55,7 @@ extern "C" {
  r_data_list,
  r_state_specification,
  r_prior,
- r_options, 
+ r_options,
  &io_manager));
 
  // Do one posterior sampling step before getting ready to write. This
@@ -88,7 +88,7 @@ extern "C" {
  << "Time used was "
  << double(current_time - start_time) / CLOCKS_PER_SEC
  << " seconds.";
- Rf_warning(warning.str().c_str());
+ Rf_warning("%s", warning.str().c_str());
  return BOOM::appendListElement(
  ans,
  ToRVector(BOOM::Vector(1, i + 1)),

src/multivariate_gaussian_model_manager.cc

@@ -100,10 +100,16 @@ namespace BOOM {
  CreateBareModel(r_data_list_or_model_object, r_prior, r_options, io_manager);
  ConditionallyIndependentSharedStateModelFactory
  shared_state_model_factory(nseries_, io_manager);
+ std::vector<Ptr<UnivParams>> residual_variance_parameters;
+ for (int i = 0; i < model_->nseries(); ++i) {
+ residual_variance_parameters.push_back(
+ model_->observation_model()->model(i)->Sigsq_prm());
+ }
  shared_state_model_factory.AddState(
  model_->state_models(),
  model_.get(),
  r_shared_state_specification,
+ residual_variance_parameters,
  "");
  shared_state_model_factory.SaveFinalState(model_.get(), &final_state());
 
@@ -137,8 +143,8 @@ namespace BOOM {
  RListIoManager *subordinate_io_manager =
  subordinate_model_io->subordinate_io_manager(i);
  StateModelFactory series_state_factory(subordinate_io_manager);
- ProxyScalarStateSpaceModel *subordinate_model =
- model_->series_specific_model(i).get();
+ ProxyScalarStateSpaceModel<MultivariateStateSpaceRegressionModel>
+ *subordinate_model = model_->series_specific_model(i);
 
  series_state_factory.AddState(
  subordinate_model, r_subordinate_state_specification);
@@ -329,7 +335,7 @@ namespace BOOM {
  }
  for (int i = 0; i < Rf_length(r_prior); ++i) {
  BOOM::RInterface::SetRegressionSampler(
- model_->observation_model()->model(i).get(),
+ model_->observation_model()->model(i),
  VECTOR_ELT(r_prior, i));
  }
 

src/state_space_student_model_manager.cc

@@ -38,8 +38,8 @@ namespace BOOM {
  RListIoManager *io_manager) {
  Matrix predictors;
  Vector response;
- std::vector<bool> response_is_observed;
  if (!Rf_isNull(r_data_list)) {
+ std::vector<bool> response_is_observed;
  // If we were passed data from R then use it to build the model.
  SEXP r_predictors = getListElement(r_data_list, "predictors");
  if (Rf_inherits(r_data_list, "bsts")) {