fuzzy-rough-learn API¶

This is the full API documentation of fuzzy-rough-learn.

Classifiers¶

`FRNN`	Implementation of Fuzzy Rough Nearest Neighbour (FRNN) classification.
`FROVOCO`	Implementation of the Fuzzy Rough OVO COmbination (FROVOCO) ensemble classifier [Rcffc929d1d8e-1].
`FRONEC`	Implementation of the Fuzzy ROugh NEighbourhood Consensus (FRONEC) multilabel classifier.

Data descriptors¶

`ALP`	Implementation of the Average Localised Proximity (ALP) data descriptor [R4d6affda87e1-1].
`CD`	Implementation of the Centre Distance (CD) data descriptor.
`EIF`	Wrapper for the Extended Isolation Forest (IF) data descriptor [R4b2ca5e1909e-1].
`IF`	Wrapper for the Isolation Forest (IF) data descriptor [R288ff40200ea-1] implemented in scikit-learn.
`LNND`	Implementation of the Localised Nearest Neighbour Distance (LNND) data descriptor [Rdc09924e667a-1]_[Rdc09924e667a-2]_.
`LOF`	Implementation of the Local Outlier Factor (LOF) data descriptor [Race47188aa49-1].
`MD`	Implementation of the Mahalanobis Distance (MD) data descriptor [Rddee2b422011-1].
`NND`	Implementation of the Nearest Neighbour Distance (NND) data descriptor, which goes back to at least [R2f1514802a03-1].
`SVM`	Wrapper for the Support Vector Machine (SVM) data descriptor [R52cd54939bff-1] with gaussian kernel, implemented in scikit-learn.

Regressors¶

FRNN

Implementation of the Fuzzy Rough Nearest Neighbour (FRNN) regressor [R889f5cf58f8d-1].

Feature preprocessors¶

Linear normalisers¶

`LinearNormaliser`	Linearly transforms all features by normalising a measure of dispersion and a measure of location, ensuring that for each feature, that measure of dispersion becomes 1 and that measure of location becomes 0.
`IQRNormaliser`	Implementation of the interquartile range (IQR) normaliser.
`MaxAbsNormaliser`	Implementation of the maximum absolute value normaliser.
`RangeNormaliser`	Implementation of the range normaliser.
`Standardiser`	Implementation of the standard deviation normaliser, or standardiser.

Other¶

`FRFS`	Implementation of the Fuzzy Rough Feature Selection (FRFS) preprocessor.
`VectorSizeNormaliser`	Rescales each instance (seen as a vector) to size 1.

Instance preprocessors¶

FRPS

Implementation of the Fuzzy Rough Prototype Selection (FRPS) preprocessor.

Other¶

Array functions¶

`div_or`	Divides x by y, replacing np.nan values with fallback.
`first`	Returns the k first values of a along the specified axis.
`greatest`	Returns the k greatest values of a along the specified axis, in order.
`last`	Returns the k last values of a along the specified axis, in reverse order.
`least`	Returns the k least values of a along the specified axis, in order.
`remove_diagonal`	Remove the diagonal from a square array.
`soft_head`	Calculates the soft head of an array.
`soft_max`	Calculates the soft maximum of an array.
`soft_min`	Calculates the soft minimum of an array.
`soft_tail`	Calculates the soft tail of an array.

Dispersion measures¶

`interquartile_range`	Distance between the first and the third quartile; range of the central half of the data.
`maximum_absolute_value`	Maximum distance from 0.
`standard_deviation`	Square root of the sum of the squared distances to the mean.
`total_range`	Distance between the smallest and largest value.

Location measures¶

`maximum`	Greatest value.
`mean`	Sum of all values divided by the number of values.
`median`	Middle value after sorting all values by size, or mean of the two middle values.
`midhinge`	Mean of the first and third quartiles.
`midrange`	Mean of the minimum and maximum.
`minimum`	Least value.

Nearest neighbour search algorithms¶

`BallTree`	Nearest neighbour search with a Ball tree.
`KDTree`	Nearest neighbour search with a KD-tree.

Parametrisations¶

`log_multiple`	Function to obtain multiples of the logarithm of other numbers.
`multiple`	Function to obtain multiples of other numbers.

T-norms¶

`goguen_t_norm`	x * y; also known as product t-norm.
`heyting_t_norm`	min(x, y); also known as Gödel or minimum t-norm.
`lukasiewicz_t_norm`	max(x + y - 1, 0)

Transformations¶

`contract`	Strictly order-preserving function from [-∞, ∞] to [0, 1] that sends -∞, -c, 0, c, ∞ to 0, 0.25, 0.5, 0.75, 1, respectively.
`shifted_reciprocal`	Order-reversing function from [0, ∞) to [0, 1] that sends x to 1/(1 + x/c).
`truncated_complement`	Order-reversing function from [0, ∞) to [0, 1] that sends x to max(0, 1 - x).

Vector size measures¶

MinkowskiSize

Family of vector size measures of the form (x1**p + x2**p + …+ xm**p)**(1/p) (if unrooted = False), or (x1**p + x2**p + …+ xm**p) (if unrooted = True), for 0 < p < ∞, and their limits in 0 and ∞..

Weights¶

`Weights`	Abstract base class for parametrisable weights functions.
`ConstantWeights`	(1/4, 1/4, 1/4, 1/4) Also known as mean weights, as they compute the unweighted mean.
`ExponentialWeights`	(8/15, 4/15, 2/15, 1/15) Exponentially decreasing weights with parametrisable base.
`LinearWeights`	(4/10, 3/10, 2/10, 1/10) Also known as additive weights.
`QuantifierWeights`	Weights that encode a regular non-decreasing quantifier [Rcc56311c3f1f-1].
`ReciprocallyLinearWeights`	(12/25, 12/50, 12/75, 12/100) Also known as inverse additive weights.