cartographer.coverers

Coverers of the filtrated space

class cartographer.coverers.HyperRectangleCoverer(intervals=10, overlap=0.5)

Covers the space using overlapping hyperectangles

Parameters:

intervals: integer or list of integers

number of intervals in each filtered space dimension, if an integer is specified the same number is used in all dimensions.

overlap: float or list of floats

fraction of overlap between hyperectangles in each space dimension, if a single float is specified the same overlap is used in all dimensions.

Methods

fit(X[, y])	Creates the space covering for the input data
fit_transform(X[, y])	Fit to data, then transform it.
get_params([deep])	Get parameters for this estimator.
overlap_matrix()	Returns a boolean array with the overlaps between space partitions
set_params(\*\*params)	Set the parameters of this estimator.
transform(X[, y])	Returns boolean array of space partition membership

fit(X, y=None)

Creates the space covering for the input data

It creates a hyperectangle covering of the multidimensional space of X.

Parameters:

X: array-like, shape=(n_samples, n_features)

Data which will be covered.

overlap_matrix()

Returns a boolean array with the overlaps between space partitions

Returns a (n_partitions, n_partitions) boolean array whose elements are true when there is overlap between the i and j partitions, only upper triangle is filled (rest is False).

Returns:

overlap_matrix: boolean array, shape=(n_partitions, n_partitions)

Boolean matrix of overlaping between partitions, only the upper triangle is filled and the rest is False.

transform(X, y=None)

Returns boolean array of space partition membership

Returns a (n_samples, n_partitions) boolean array whose elements are true when the sample (row) is a member of each space partition (column). This will be used to filter in the clustering space.

Parameters:

X: array-like, shape=(n_samples, n_features)

Data which will be partition in hyperectangles.

Returns:

m_matrix: boolean array, shape=(n_samples, n_partitions)

Boolean matrix of sample membership to each partition