                        Changes in version 0.2.0                        

  - Added new method for fuzzy number ordering as metd = 5 in functions
    GauNBFuzzyParam, ExpNBFuzzyParam, and PoiNBFuzzyParam.
  - Significant performance optimization across all classifiers:
    vectorized predictions using native matrix operations and max.col
    for substantial speedups.
  - Fixed discrete variables validation (verifyNumbersFunction) ensuring
    compatibility with R 4.2+ and resolving missing argument errors in
    DWFuzzyHipergeometricNaiveBayes, FuzzyBinomialNaiveBayes,
    FuzzyHipergeometricNaiveBayes, and FuzzyPoissonNaiveBayes.
  - Fixed mathematical consistency in FuzzyHipergeometricNaiveBayes by
    applying joint membership once to the class score, consistent with
    other models.
  - Fixed prior class probability multiplication in FuzzyGeoNaiveBayes
    applied once per class.
  - Cleaned up package dependencies: removed unused dependencies (dplyr,
    tibble, tidyr, rlang, purrr) from Imports.
  - Fixed class/label misalignment: parameters were estimated in the
    order the classes appear in the data while predictions were labelled
    with the sorted factor levels, so predictions were wrong whenever
    the training data were not sorted by class. All classifiers now use
    the factor levels (get_classes()) for estimation and prediction, and
    the columns of type = "matrix" follow the levels.
  - FuzzyGammaNaiveBayes / DWFuzzyGammaNaiveBayes: exact Gamma maximum
    likelihood estimates (the previous optimisation was bounded by
    max(x), which truncated the shape/scale).
  - FuzzyBetaNaiveBayes: removed the artificial upper bound (10) on the
    Beta shapes; the method of moments is used as starting value.
  - FuzzyGeoNaiveBayes: probability mass function consistent with the
    estimator (dgeom, support 0, 1, 2, ...); memberships use the same
    intervals in training and prediction, empty intervals (and values
    outside the training range) receive 0.001.
  - FuzzyBinomialNaiveBayes: closed-form method of moments with n >=
    max(x).
  - GauNBFuzzyParam / ExpNBFuzzyParam: confidence intervals (alpha-cuts)
    use the class sample size instead of the total sample size.
    ExpNBFuzzyParam ranks densities (as the Gaussian and Poisson
    versions), since the ranking methods assume non-negative fuzzy
    numbers.
  - PoiNBFuzzyParam: alpha-cuts are now nested (they shrink as alpha
    increases).
  - Ranking methods (metd = 1..5) vectorized; the closed form of the Q
    distance replaces one eigen decomposition per observation.
    doParallel, foreach and parallel are no longer needed; cores is kept
    for backward compatibility.
  - FuzzyTriangularNaiveBayes / FuzzyTrapezoidalNaiveBayes: densities
    outside the support are floored at 1e-10, avoiding ties decided by
    class order.
  - FuzzyNaiveBayes: the arguments m and Pi are now passed to the
    estimators (they were ignored).

                 Changes in version 0.1.7 (2025-09-03)                  

  - Add the "Fuzzy Hipergeometric Naive Bayes classifier" as
    FuzzyHipergeometricNaiveBayes
  - Add the "Double Weighted Fuzzy Hipergeometric Naive Bayes
    classifier" as DWFuzzyHipergeometricNaiveBayes
  - Fix: memberships functions. Now they work correctly with dataframes

                 Changes in version 0.1.6 (2024-02-14)                  

  - Add the "Fuzzy Rule-based System" with some functions
  - Fix: FuzzyNaiveBayes to one column in the data
  - Fix: "membership" to dataframe with one column in the data
  - Fix: "Frequency" to discrete distributions

                 Changes in version 0.1.5 (2023-08-17)                  

  - Add the "double weighted fuzzy gamma naive bayes classifier" as
    DWFuzzyGammaNaiveBayes
  - Add the "geometric fuzzy gamma naive bayes classifier" as
    GeometricFuzzyGammaNaiveBayes
  - Add the "Fuzzy Bayes Rule classifier" as FuzzyBayesRule
  - Remove the ordering method for functions with fuzzy parameters
    metd=4.

                 Changes in version 0.1.4 (2023-03-07)                  

  - Added a new ordering method for functions with fuzzy parameters:
    GauNBFuzzyParam, ExpNBFuzzyParam,GamNBFuzzyParam and
    PoiNBFuzzyParam. The form can be used considering metd = 4.

                 Changes in version 0.1.3 (2022-11-16)                  

  - Update in parameters estimation in function
    FuzzyTriangularNaiveBayes.
  - Update predict has been optimized to functions FuzzyGammaNaiveBayes.
  - Update functions of GamNBFuzzyParam and GauNBFuzzyParam.
  - predict function has been optimized to FuzzyGammaNaiveBayes,
    FuzzyBetaNaiveBayes,
    FuzzyBinomialNaiveBayes,FuzzyExponencialNaiveBayes,FuzzyTriangularNaiveBayes,FuzzyPoissonNaiveBayes

                 Changes in version 0.1.2 (2022-08-25)                  

New features

  - Changed \() by function(.) in function FuzzyNaiveBayes, line 262.
  - Changed description in FuzzyClass Documentation

                 Changes in version 0.1.1 (2022-05-27)                  

New features

  - Estimation new form in function FuzzyGammaNaiveBayes.
