Skip to content

Latest commit

 

History

History
117 lines (88 loc) · 5.36 KB

NEWS.md

File metadata and controls

117 lines (88 loc) · 5.36 KB
Raw

DAISIEutils 1.6.2

  • Prevent parameter values above upper bound being passed to the relaxed rate model in run_daisie_ml() (#34).
  • Lower the value given as initial parameter value when estimated as infinite for the relaxed rate model in run_daisie_ml().

DAISIEutils 1.6.1

  • Migrate from now defunct Peregrine HPCC to the new Hábrók HPCC
  • Prevent Inf from being passed to the relaxed rate model in run_daisie_ml() (#33)

DAISIEutils 1.6.0

  • The version of R and DAISIE are incremented to 4.2 and 4.4.0, respectively
  • The relaxed-rate DAISIE model now has an initial DAISIE optimisation to get better initial conditions (in run_daisie_ml())
  • Removed an old documentation section causing warning

DAISIEutils 1.5.0

  • Allow relaxed-rate DAISIE ML models.
    • This requires new a argument for run_daisie_ml() and setup_model(): par_upper_bound, which sets the upper limit of the integration of a relaxed parameter. This defaults to Inf in the R function and shell scripts, which is no upper bound of integration for the relaxed-rate DAISIE model. This parameter is ignored when using the standard constant-rate case (i.e., not relaxed-rate).
  • Allow 2 type DAISIE ML analyses, handled by run_daisie_2type_ml() and adjacent function setup_2type_model(). Similarly add required R run_daisie_2type_ml.R script and shell scripts submit_run_daisie_2type_ml.sh and submit_run_daisie_2type_ml_long.sh to run said analyses in an HPCC.
  • Package depends on CRAN DAISIE release instead of GitHub repository. Now requires DAISIE >= v4.3.1 to ensure latest ML related bugfixes are used.
  • Add new tests covering new cases.
  • Add Rampal Etienne's details to zenodo release.

DAISIEutils 1.4.0

  • Add new argument res to change resolution in DAISIE::DAISIE_ML_CS(). Default values allows for backwards compatibility in functions and job scripts for Peregrine.

DAISIEutils 1.3.0

  • Can now extract a single data set (or replicate) from a data set that stores several within the same list.
  • Add support for non-oceanic models (which can be chosen in the relevant functions and start with nonoceanic in relevant functions).

DAISIEutils 1.2.1

  • Correct .zenodo.json for automatic release and archiving on Zenodo.

DAISIEutils 1.2.0

  • Reworked all reference test file infrastructure to use tempdir().
  • Added results_dir argument to functions that load and/or write to the file system to allow the user to specify a custom directory appropriate for their environment. The default, NULL maintains previous behaviour, i.e., saves and loads from a folder results/ in the root of the working directory.
  • Removed is_daisie_data() as it was incomplete and seldom used. May be ported from other packages in the future.
  • Rework create_output_folder() to only handle directory creation. The file path generation is now handled by create_results_dir_path() assuming previous functionality with new added flexibility via the results_dir argument as described above.
  • Add alternative (lower) CES rates to run_daisie_ml() and setup model to allow certain datasets to be begin estimation from valid parameters.
  • Renamed argument data to daisie_data for consistency with more recent DAISIE related packages and to avoid conflicts with base R's data().
  • Add functions to plot bootstrap results to check model goodness of fit: plot_bootstrap_results(), summarize_bootstrap_results() adding to plot_sim_metrics() which now is split into calc_sim_metrics().
  • Add setup_std_pars2() to generate common pars2, useful for development within 'DAISIE'.
  • run_daisie_ml() can now return it's output to session rather than saving to file by setting results_dir to NA.
  • run_daisie_ml() uses lsodes as default methode, in line with 'DAISIE'.
  • Style entire package with 'styler'.
  • Require 'DAISIE' v4.2.1.
  • No longer depend on private packages, to ensure package can be accessed by more users.
  • Due to new plot functions, depend on 'ggplot2'and 'cowplot'.
  • Added upload_results.R and upload_results.sh to upload to Google drive directly from Peregrine.
  • Added .zenodo.json with metadata for automatic Zenodo releases.

DAISIEutils 1.1.0

  • choose_best_model() correctly handles results where no model was estimated successfully, and returns NA appropriately.

  • sensitivity() now works correctly regardless of the number of parameters used to estimate the chosen models. This means relaxed-rate models and any model fitting that returns results with more than the base DAISIE parameters is accommodated.

  • sensitivity() no longer saves to file and instead returns results to the environment.

  • Improved sensitivity() documentation.

  • Depend on and install DAISIE v4.0.2.

DAISIEutils 1.0.0

  • Complete overhaul of package.

  • Add run_daisie_ml() to fit DAISIE models on DAISIE datasets. Returns model fitting results and BIC value.

  • Add bootstrap_lr() to conduct a likelihood ratio bootstrap test on two DAISIE models.

  • Add bootstrap() to conduct goodness of fit bootstrapping test.

  • Add sensitivity() to calculate the sensitivity of the model to two alternative data sets.

DAISIEutils 0.0.0.9001

  • Create package skeleton.
  • Add print_main_header().
  • Use default_params_doc.R to document package.
  • Write README.md stub.
  • Add tests and coverage.
  • Added a NEWS.md file to track changes to the package.