MFF 0.2.5

• Fixed parallel tune.mff() grid searches by making all internal functions required by worker processes available in their execution environments.
• mff() now requires at least two clusters/functions, consistently with the search space used by tune.mff().
• Added unit tests for the mff() cluster-count boundary and equality of sequential and parallel k-means tuning results under a common seed.

MFF 0.2.4

• Added plot.mff(type = "weight_heatmap") for a labelled heatmap of the complete candidate-model by meta-function weight matrix.
• Added an optional distance-preserving PCA representation for clustering candidate prediction profiles. This reduces the clustering space to at most the number of candidate models minus one and makes GK clustering practical when the validation set is much larger than the candidate set.
• Strengthened validation of prediction matrices, response vectors, clustering parameters, tuning grids, and parallel execution arguments.
• predict.mff() now verifies candidate-model identities and column order before applying fitted weights.
• predict.mff(type = "best") now reports an informative error when no validation-selected weight is available, instead of implicitly returning all meta fuzzy functions.
• evaluate() now rejects mismatched, empty, malformed, and non-finite inputs.

MFF 0.2.3

• Added an end-to-end vignette demonstrating the boot.train()–tune.mff()–predict()–evaluate() workflow and the separation of validation-based selection from final test evaluation.
• Added a testthat unit-test suite covering the public interface, all four membership-generation methods, input validation, prediction-matrix dimensions, and sequential/parallel bootstrap reproducibility.
• Reworked examples to use lightweight prediction matrices and disabled parallel execution explicitly in examples and the vignette.
• Limited XGBoost and LightGBM inside model.train() to one thread to avoid oversubscribing CRAN check machines.
• Added print(), summary(), and plot() methods for fitted and tuned MFF objects.
• Extended plot.mff() with validation-safe test score, observed-versus-predicted, and test-series visualizations. Test plots use the validation-selected function by default and never reselect a function using test performance.
• Made the computationally intensive model.train() example interactive-only so that CRAN example checks do not initialize several external learner libraries merely to demonstrate the convenience function.