The hardware and bandwidth for this mirror is donated by METANET, the Webhosting and Full Service-Cloud Provider.
If you wish to report a bug, or if you are interested in having us mirror your free-software or open-source project, please feel free to contact us at mirror[@]metanet.ch.

AddiVortes News

AddiVortes 0.6.9

• Fixed a heap-buffer-overflow in propose_internal() when proposing a Swap Dimension move on spherical covariates. The period-shift check indexed the tessellation’s active-dimension vector by the newly drawn covariate index rather than testing that covariate against members, reading past the end of the vector whenever the tessellation used fewer dimensions than covariates. This caused AddressSanitizer aborts when rebuilding the spherical vignette under CRAN’s SAN checks.
• Removed an unused local variable in log_acceptance_components() that triggered a significant compiler warning (-Wunused-variable) and caused CRAN’s “checking whether package can be installed” WARNING.

AddiVortes 0.6.8

• Corrected three Metropolis-Hastings structural-move bugs in log_acceptance_components():
  • add-centre no longer applies an extra 1/(b+1) penalty that cancelled against set multiplicity and thinned the cell-count prior;
  • add/remove-dimension no longer double-count the uniform covariate pick already present in the set prior;
  • all six Appendix B boundary proposal adjustments are now applied (the three reverse x2 partners had been missing).
• With these corrections, the default LambdaRate = 25 matches the stated Poisson prior (~26 cells per tessellation). Previously the thinned ratio produced about 5 cells at that default.

AddiVortes 0.6.7

• AddiVortes can now calculate distances between categorical covariates without needing to apply one-hot encoding, by specifying cat.onehot = FALSE in the main AddiVortes() function. The distance calculated when one-hot encoding is not used is Eskin distance (Eskin et al., 2002); other measures may be added to the framework in the future.

• A bug in the C code, where indices were not referenced correctly in distance calculations, has been fixed in add_vortes_mcmc_cpp(): distance calculations across multiple different covariate types and dimensions now agrees with the expected values.

• Due to the development of the package, the propose_tessellation_cpp() helper function is no longer used; to avoid code duplication, it has been removed from this version.

AddiVortes 0.6.6

• Fixed out-of-sample prediction so that cell assignments match the in-sample fit. cellIndices() now always maps tessellation centre columns to their global covariate positions (previously the remapping was skipped when a tessellation used every covariate, silently mismatching coordinates whenever the active dimensions were in a permuted order). predict() now derives its spherical column mask from the reordered/augmented metric so spherical coordinates are left on their original radian scale even when covariateStructure_internal() reorders covariates. Predictions on the training data now reproduce the in-sample fitted values, including for spherical and mixed-metric models.

AddiVortes 0.6.5

• Metropolis-Hastings acceptance ratios have been further simplified in the C++ MCMC implementation.
• Removed legacy internal R functions for MCMC steps (acceptance probabilities, tessellation proposals, residual aggregation, and mu/sigma sampling) that are now handled entirely in C++.

AddiVortes 0.6.4

• Corrected the log-likelihood trace in traceplots() so it now shows the retained-state log-likelihood component at the end of each MCMC iteration, rather than the average proposal log-likelihood-ratio component evaluated during the sweep.

AddiVortes 0.6.3

• traceplots() now includes the burn-in period in its MCMC traces and colours burn-in sections black.
• Replaced the error standard deviation panel in traceplots() with the log-likelihood component used in tessellation acceptance ratios.

AddiVortes 0.6.2

• Added traceplots() for AddiVortes objects, showing trace plots for average tessellation centres, the standard deviation of tessellation centres, average active dimensions, and the error standard deviation.

AddiVortes 0.6.1

• AddiVortes now supports covariates drawn from multiple subspaces of the same type and allows future support for more complex categorical covariate distance metrics; AddiVortes now takes a members argument as well as the metric argument to identify different covariate subspaces.
• Distance calculation has been streamlined and modularised via data preprocessing in covariateStructure and associated changes in C++ code.
• Package dependency has been updated to explicitly require C++20 compilation.

AddiVortes 0.6.0

• Consolidated model fitting into a single C++ call (addi_vortes_mcmc_cpp). The complete MCMC loop — sigma sampling, tessellation proposals, cell-index assignment, residual aggregation, acceptance-probability evaluation, and mu sampling — now runs entirely in C++, eliminating the per-iteration R↔︎C++ overhead that was present in previous versions.
• The AddiVortes class and all associated S3 methods are unchanged.

AddiVortes 0.5.4

• Added data to package to help with vignettes and testing.

AddiVortes 0.5.3

• Added vignette “Modelling Spherical Data with AddiVortes” (by Andy Iskauskas and John Paul Gosling) demonstrating the use of great-circle distance via metric = "S", the coordinate convention (latitude/longitude in radians), and a comparison with the Euclidean metric on synthetic globe data.
• Added vignette “Using Categorical Covariates with AddiVortes” explaining automatic one-hot encoding (d categories → d-1 binary indicators), the catScaling parameter, the catEncoding metadata field, and handling of unseen category levels at prediction time.

AddiVortes 0.5.2

• Added support for categorical covariates via automatic one-hot encoding. Character and factor columns are converted to d-1 binary indicator variables (first level as reference), controlled by the new catScaling parameter.
• predict.AddiVortes() now accepts data.frame inputs when the model was trained with categorical covariates, applying the stored encoding automatically.
• Unseen categories at prediction time are mapped to the reference level (all zeros).
• Encoding metadata stored in the catEncoding field of the fitted model object.
• Added comprehensive tests for categorical covariate support.
• Added second example demonstrating categorical covariates with a larger training (n=200) and test set (n=50).

AddiVortes 0.5.1

• Reinstated C++ speed-ups for new distance metrics
• Restructured testthat tests

AddiVortes 0.5.0

• Added compatibility for purely spherical input data and mixed Euclidean and Spherical inputs.

AddiVortes 0.4.12

• Added comprehensive test suite for AddiVortes class methods syntax validation.
• Tests cover input validation and error handling for all S3 methods (print, summary, predict, plot).

AddiVortes 0.4.11

• Renamed class AddiVortesFit to AddiVortes to maintain consistency with the package name.
• Updated all S3 methods (print, summary, predict, plot) to use the new class name.
• Updated constructor function from new_AddiVortesFit to new_AddiVortes.

AddiVortes 0.4.10

• Fixed variance/standard deviation mismatch in tessellation proposal step.
• Fixed typo in variable names and added more camelCase.

AddiVortes 0.4.9

• Improved legends in plot.AddiVortesFit().
• Fix typos in package description.

AddiVortes 0.4.8

• Initial CRAN release of AddiVortes package.
• Fixed installation failure on r-devel-linux-x86_64-fedora-clang:
  • Added missing <cstring> header for memcpy() function in C++ code
  • Clang compiler requires explicit inclusion of standard library headers

AddiVortes 0.4.7

• Fixed DESCRIPTION file for CRAN compliance:
  • Removed non-standard ‘Keywords’ field
  • Removed redundant ‘Author’ and ‘Maintainer’ fields (now auto-derived from Authors@R)
• Resolved R CMD check NOTEs for DESCRIPTION meta-information

AddiVortes 0.4.6

• Linting and formatting improvements.
• Final preparations for CRAN submission.

AddiVortes 0.4.5

• Fixed bug in tessellation proposal when number of covariates equals number of selected dimensions.
• Add Dimension (AD) modification now properly checks if all covariates are already selected before attempting to add a new one.
• Added comprehensive test suite for small covariate counts (1, 2, and 3 covariates).

AddiVortes 0.4.4

• Fixed test automation bug stemming from too many cores being assumed available.

AddiVortes 0.4.3

• Cleaned up tests folder for CRAN submission preparation.
• Removed TestSuite.R, CodeProfiler.R, and TestHelper.R from tests directory.
• Incorporated relevant tests from TestSuite.R into testthat framework.
• Added test for thinning parameter functionality.
• Only testthat.R remains in tests folder alongside the testthat directory per CRAN requirements.

AddiVortes 0.4.2

• Added prediction interval support to predict.AddiVortesFit() function with new interval parameter.
• Fixed bug where posteriorSigma was not stored in model objects.
• Prediction intervals now available alongside confidence intervals, similar to lm predict function.
• Updated tests to use named access instead of numeric indexing for improved robustness.

AddiVortes 0.4.1

• Added parallel processing to predict function.
• Removed unnecessary square-root function from KD-tree search.

AddiVortes 0.3.3

• Added warning for situation where covariates exceed observations.

AddiVortes 0.3.2

• Improved progress bar output to give more information during MCMC sampling and prediction processes.
• Removed outputted progress bars from vignette examples to reduce clutter.

AddiVortes 0.3.1

• Enhanced SEO and discoverability with comprehensive keyword optimization across package documentation
• Added formal Keywords field in DESCRIPTION for better search indexing
• Updated README.md with clear positioning as BART alternative
• Optimized vignette titles for machine learning and Bayesian regression search terms
• Enhanced package documentation with machine learning focus

AddiVortes 0.3.0

• Implemented nearest neighbour search in C++.
• Removed dependency on FNN package.

AddiVortes 0.2.5

Package Features

• Added progress bars to MCMC sampling functions and subsequent prediction functions for better user feedback during long computations.

AddiVortes 0.2.4

CRAN Submission Preparation

• Removed compiled object files from package source
• Added examples to main AddiVortes() function
• Updated DESCRIPTION file for CRAN compliance:
  • Added Author and Maintainer fields
    
  • Added R version dependency (>= 3.5.0)
  • Updated Date field
  • Fixed grammatical error in Description
• Updated .Rbuildignore with standard exclusions
• Enhanced documentation with examples for key functions

Package Features

• Implements Bayesian Additive Voronoi Tessellation models
• Non-parametric regression with tessellation-based approach
  
• Posterior sampling via backfitting algorithm
• Prediction and visualization methods for fitted models
• Comprehensive test suite and vignettes

These binaries (installable software) and packages are in development.
They may not be fully stable and should be used with caution. We make no claims about them.