NEWS

---

spca 1.1.4 (2026-09-23)

• spca's parameter fixed_index_list now accepts same index in different components it also allows to require that only some components have fixed indices.
• created new class pca for pca() output with its own methods.
• renamed pca plotting methods to scree_plot and mp_qqplot(). Old functions kept as deprecated
• scree_plot produces x-axis with integer tick-marks
• Heat map has low = red and high = blue
• eliminated help entries in index for deprecated functions
• improved symmetry check for gramian input
• fixed minor bugs in methods, reinforced checks for cor and n_obs in mp_qqplot and pca
• reviewed and harmonized vignettes with JSS article

spca 1.1.3 (2026-08-31)

Renamed loadings to weights

• objects now return weights, as documented.
• same changes in C++ code but no need of backcompatibility
• preserved backward compatibility by retaining the old function names as
  deprecated wrappers for the new S3 methods
• preserved compatibility with older spca objects containing loadings-named elements through internal accessors

Converted functions to methods and added new ones

• changed function aggregate_by_group_spca() to method aggregate_by_group()
• changed function show_contributions_spca() to method show_weights()
• added method show_correlations()
• changed function change_weights_sign_spca() to method change_sign()
• changed functions spca_screeplot() and wachter_qqplot() to methods screeplot_spca() and qqplot_spca() -- first argument changed from eigenvalues to pca_obj in qqplot_spca(), arguments n and p changed to n_vars and n_obs
• added the method show_correlations to visualize the correlations between sPCs and their correlation with the original PCs.
• pca() output takes class c("spca_pca", "spca", "list").

Other cosmetic changes

• summary now returns cor_Pcs by default
• removed ggplot warnings aboout missing data

spca 1.1.2

• fixed cor_with_pcs sign in change_loadings_sign_spca

spca 1.1.1 (2026-07-10)

• fixes for release
• removed windows only compiler flags
• removed references to deceased website
• various minor changes

spca 1.1.0

Major changes

• Added projection SPCA (method = "pspca"), including a new C++ implementation.
• Added a dedicated C++ backend for fat matrices, where the number of variables exceeds the number of observations.
• Redesigned the C++ tall-matrix engine for improved speed, memory use, and maintainability.
• Refactored the C++ codebase so large objects are passed by reference where appropriate.
• Added try/catch wrappers around the main C++ entry points to improve error reporting from compiled code.

R interface

• Reworked the main spca() interface while preserving the core LS-SPCA workflow.
• Rewrote plot.spca() with improved bar plots, circular plots, heat maps, grouping support, and plotting controls.
• Rewrote new_spca() to make externally computed sparse loading matrices easier to store and inspect as spca objects.
• Improved printing and summary methods for spca objects.
• Improved comparison tools for multiple sparse PCA fits.
• Added or revised helper functions for grouped summaries, diagnostics, and validation.

Reliability and testing

• Added a testthat test suite.
• Fixed multiple issues in the R interface and C++ backend.
• Improved input validation and error messages.
• Fixed edge cases affecting sparse loadings, component summaries, and variance-explained diagnostics.

Documentation

• Rewrote and expanded the package documentation.
• Rewrote the vignettes for the revamped package.
• Updated examples to reflect the current interface and available LS-SPCA methods.
• Added documentation for projection SPCA and the tall/fat backend distinction.

spca 1.0.0

Major rewrite

• First full release of the revamped spca package.
• Replaced the previous computational core with a new C++ backend.
• Added separate computational paths for tall and fat data matrices.
• Added projection SPCA as a new LS-SPCA method.
• Updated the package interface to support the redesigned backend.

Package structure

• Reorganized internal R and C++ code.
• Improved consistency of returned spca objects.
• Updated exported functions and methods to use a common object structure.
• Added tests and revised examples for the new implementation.

spca 0.9.0

Development release

• Introduced the new C++ tall-matrix engine.
• Began migration of computational routines from the earlier implementation to the redesigned backend.
• Added preliminary support for projection SPCA.
• Improved memory handling in compiled code.
• Updated internal validation routines.

spca 0.8.0

Development release

• Reworked plotting methods for spca objects.
• Added improved graphical comparison of loadings and contributions.
• Added support for grouped variables in summaries and plots.
• Revised helper functions used by examples and vignettes.
• Improved documentation of sparse loadings, contributions, VEXP, CVEXP, and component correlations.

spca 0.7.0

Development release

• Began the package revamp following version 0.6.0.
• Reorganized the R interface around fitted spca objects.
• Improved print and summary output.
• Added early validation and diagnostic improvements.
• Fixed several bugs from the previous CRAN release.

spca 0.6.0 (2014-12-16)

• Previous CRAN release.