stag.kde.ExactGaussianKDE Class Reference

Description

A data structure for computing the exact Gauussian KDE.

This data structure uses a brute-force algorithm to compute the kernel density of each query point.

The time complexity of initialisation with \(n\) data points is \(O(1)\). The query time complexity is \(O(m n d)\), where \(m\) is the number of query points, and \(d\) is the dimensionality of the data.

Public Member Functions
def	__init__ (self, stag.utility.DenseMat data, float a)
	Initialise the data structure with the given dataset and Gaussian kernel parameter \(a\).
Union[float, np.ndarray]	query (self, Union[stag.utility.DenseMat, stag.data.DataPoint] q)
	Calculate the exact kernel density estimates for the given query points.

Public Attributes
	internal_kde

Constructor & Destructor Documentation

__init__()

def stag.kde.ExactGaussianKDE.__init__	(		self,
		stag.utility.DenseMat	data,
		float	a
	)

Initialise the data structure with the given dataset and Gaussian kernel parameter \(a\).

The initialisation time for this data structure is \(O(1)\).

Parameters

data
a

Member Function Documentation

query()

Union[float, np.ndarray] stag.kde.ExactGaussianKDE.query	(		self,
		Union[stag.utility.DenseMat, stag.data.DataPoint]	q
	)

Calculate the exact kernel density estimates for the given query points.

The parameter q can be either a stag.data.DataPoint object to query one data point, or a stag.utility.DenseMat matrix with the query points as rows in order to query many data points.

For querying many data points, passing the queries as a DenseMat will be more efficient.

Parameters

q	the query data point(s)

Returns

the kernel densities for the given query point(s)

Member Data Documentation

internal_kde

stag.kde.ExactGaussianKDE.internal_kde