SCE 1.1.3

Breaking changes (API style)

• Per common R conventions, all exported function names now use lowercase (including main constructors sce() and sca()).
• S3 class names are now "sce" and "sca" (previously "SCE" and "SCA"). Method names follow (print.sce, predict.sca, etc.).
• All function arguments now use lowercase / snake_case. Renamed arguments include:
  • Training_data -> training_data, Testing_data -> testing_data
  • X -> x, Y -> y, X_sample -> x_sample
  • Nmin -> nmin, Ntree -> ntree
  • Predictors -> predictors, Predictant -> predictant, Simulations -> simulations
  • OOB_weight -> oob_weight
  • Internal helpers: Max_merge_iter -> max_merge_iter, Weak_L -> weak_l
  • Affected functions: sce(), sca(), rfe_sce(), model_simulation(), sca_tree_predict(), sce_model_evaluation(), sca_model_evaluation(), wilks_importance(), S3 methods predict.*, importance.*, evaluate.*, etc.
• Renamed functions include (non-exhaustive): model_simulation(), sca_tree_predict(), sce_model_evaluation(), sca_model_evaluation(), wilks_importance(), sca_importance(), rfe_sce(), plot_rfe(), and internal helpers such as sce_prediction(), training_prediction(), oob_validation(), gof(), nse_equation(), kge_equation(), inference().
• Saved .RDS / .RData models created with older versions store the previous class names; assign new classes before using S3 methods, e.g. class(obj) <- "sce" or class(obj) <- "sca" as appropriate.

Bug fixes

• sce(parallel = TRUE) no longer fails on single-core machines (or when Ntree == 1); it now falls back to sequential execution instead of leaving the result vector undefined. The cluster is also released via on.exit() so a worker error no longer leaks the cluster.

Documentation

• Updated README.md, tutorial script, and manual pages to match the new function and dataset names.

SCE 1.1.2

New Features

• Added digits argument for variable importance outputs
  • Wilks_importance() and SCA_importance() now round Relative_Importance to digits decimal places (default: 2)
  • S3 methods importance.SCE() and importance.SCA() accept and forward the digits argument

Documentation

• Updated man/importance.Rd to document the digits parameter and default
• Updated README examples to demonstrate importance(..., digits = 2)

Compatibility

• Backward compatible; only changes displayed precision of importance values. Adjust digits to control rounding

SCE 1.1.1

New Features

• Added parameter validation warnings to evaluate() S3 methods
  • evaluate.SCA() now automatically retrieves predictants from the model object (no need to specify)
  • evaluate.SCA() now specifically warns if Training_data is provided (not needed for SCA evaluation)
  • evaluate.SCE() now automatically retrieves predictants from the model object (no need to specify)
  • Both methods now warn if extra parameters beyond required ones are provided
• Enhanced user experience by providing clear guidance on correct parameter usage

SCE 1.1.0

New Features

• Added S3 class system for SCE and SCA objects with dedicated methods
• Implemented print(), summary(), predict(), importance(), and evaluate() methods for S3 objects
• Added Plot_RFE() function for visualizing recursive feature elimination results
• Enhanced backward compatibility - functions now work with both S3 objects and legacy list formats

Improvements

• Improved parallel processing with proper function exports to cluster
• Enhanced error handling and input validation
• Updated documentation with comprehensive examples using S3 methods
• Consolidated documentation files for better organization

Bug Fixes

• Resolved S3 method dispatch problems
• Corrected parameter order in Model_simulation() and SCA_tree_predict() functions

Documentation

• Updated all .Rd files to use S3 methods in examples
• Added comprehensive README with installation and usage instructions
• Improved function documentation with better examples
• Added cross-references between related functions

Technical Changes

• Added proper NAMESPACE configuration with S3 method registrations
• Imported required graphics functions (legend, lines) for plotting
• Enhanced package structure for better maintainability

mirror server hosted at Truenetwork, Russian Federation.