NEWS.md

In jtimonen/lgpr: Longitudinal Gaussian Process Regression

lgpr 1.2

1.2.4

• Start using rstan (>= 2.26.0) and the new array syntax.

1.2.1-1.2.3

• Internal non-functional changes.
• Use rstantools >= 2.3.1 which is needed for CRAN.

1.2.0

• Update installation process.
• Add @andrjohns as contributor.

lgpr 1.1

1.1.5

• Add more imports to NAMESPACE per a rstan developer's recommendation.

1.1.4

• New vignette about mathematical description of models.
• Prediction and function posterior computation at P points where P is larger than number of data points should be now much faster and take less memory, as P x P matrices are not computed.

1.1.3

• First CRAN release.

1.1.1-1.1.2

• Only small patches in documentation etc. in order to conform to CRAN policies

1.1.0

• Adds prior_pred() for prior predictive sampling and sample_param_prior() for sampling from the parameter prior.
• Adds read_proteomics_data() function.
• Relax data type checking, to require that they only inherit from factor or numeric. Allow also tibbles and data.tables to be passed as data.
• Adds more methods for lgpfit and lgpmodel objects, see their documentation.
• Lot of improvements internally. Kernel computations in functions like pred() should take a lot less memory now. Two separete main Stan models now. One for latent GP (signal where f is sampled) and other for GP with marginalized f.
• Improved documentation.
• Improve verbose messages to user.

lgpr 1.0

1.0.13

• Fix bug that ignored the group_by argument in get_teff_obs() and caused at least new_x() to not work if the subject identifier variable was called something else than id (see issue #22).

1.0.12

• Add more informative error message if trying to specify a model like y ~ age + id | age, which should be y ~ age + age | id, i.e. the continuous covariate on the left of | and categorical on the right.
• New startup message that prints also rstan version
• Update citation information

1.0.11

• Add the c_hat_pred argument to pred(), to be used when f has been sampled and c_hat is not constant. Previously, c_hat = 0 was used in all prediction points, which did not make sense in all cases.

1.0.10

• Allow setting group_by = NA in plot_pred(), plot_components() and new_x() to avoid grouping in plots.
• Allow setting color_by as the same factor as group_by.
• Fix bug which caused an error when trying to define a separate prior for parameters of the same type.

1.0.9

• Internal change for more effective computation of function (component) posterior variances.

1.0.8

• Add option do_yrng which controls whether to do draws from the predictive distribution. This was previously always done if sample_f was TRUE. That is now considered a bug because it is unnecessary work if the y_rng draws are not needed. So the default is now do_yrng = FALSE, since do_yrng = TRUE can cause errors with the negative binomial model due to numerical problems (see here). These problems should be addressed in a future release to allow more stable prior and posterior predictive sampling.

1.0.7

• Small documentation update.

1.0.6

• Fix bug in get_pred(), which was caused by not adding the GP mean to the sampled signal. This was causing postprocessing functions like relevances() and plot_pred() to give erroneous results if the GP mean was not a vector of zeros and sample_f = TRUE.
• Small edits in documentation and verbose information messages.

1.0.5

• Make plot_pred() work with any response variable name (fixes issue #12).
• Avoid adding ggplot2::color_scale_manual() if number of colors > 5 (fixes issue #11).

1.0.4

Edit type checking to work more generally on all systems (fixes issue #5).

1.0.3

Fix CITATION to point to new preprint.

1.0.2

Added RcppParallel dependency explicitly.

1.0.1

Added warning if using default prior for input warping steepness.

1.0.0

New features

• More general modeling options, allowing more mixing of different types of kernels/options
• Prior and posterior predictive checks using ppc(), which interfaces to bayesplot.

Changes and improvements

• Formula syntax where | indicates interaction terms.
• Alternative advanced formula syntax with gp(), gp_warp(), zerosum() etc.
• Beta binomial observation model.
• Categorical covariates must now be specified as factors in data, and don't have to be numeric.
• Component relevance assessment is now separated from model fitting into the relevances() function and selection into select().
• Easier prior specification with normal(), log_normal(), student_t() etc.
• Better prediction and plotting functionality with get_pred(), pred(), plot_pred(), and plot_f().
• Extensive argument checking (see check_positive_all() etc.) to give users informative error messages

Automated testing

• Thorough unit tests using test_that.
• C++ versions of the Stan model functions are now exposed to package namespace and also tested.

lgpr 0.33

• First release.

lgpr < 0.33

• Earlier development versions.

jtimonen/lgpr documentation built on Oct. 12, 2023, 11:13 p.m.