distancematrix.generator.znorm_euclidean

Module Contents

Classes

ZNormEuclidean	Class capable of efficiently calculating parts of the z-normalized distance matrix between two series,
BoundZNormEuclidean	Helper class that provides a standard way to create an ABC using

class distancematrix.generator.znorm_euclidean.ZNormEuclidean(noise_std=0.0, rb_scale_factor=2.0)

Bases: distancematrix.generator.abstract_generator.AbstractGenerator

Class capable of efficiently calculating parts of the z-normalized distance matrix between two series, where each entry in the distance matrix equals the euclidean distance between 2 z-normalized (zero mean and unit variance) subsequences of both series.

This generator can handle streaming data.

Subsequences with standard deviation <= 1e-6 will be treated as flat sequences to avoid problems with numerical stability.

prepare_streaming(self, m, series_window, query_window=None)

Create a bound generator that supports streaming data. The generator will need to receive data before any distances can be calculated.

Parameters

• m – the size of the subsequences used to calculate distances between series and query

• series_window – number of values to keep in memory for series, the length of the horizontal axis of the distance matrix will be equal to (series_window - m + 1)

• query_window – number of values to keep in memory for query, the length of the vertical axis of the distance matrix will be equal to (query_window - m + 1), or None to indicate a self-join.

Returns

a bound generator that supports streaming

prepare(self, m, series, query=None)

Create a bound non-streaming generator for the given series and query sequences.

Parameters

• m – the size of the subsequences used to calculate distances between series and query

• series – 1D array, used as the horizontal axis of a distance matrix

• query – 1D array, used as the vertical axis of a distance matrix, or None to indicate a self-join

Returns

a bound generator

class distancematrix.generator.znorm_euclidean.BoundZNormEuclidean(m, series, query, self_join, noise_std, series_mu, series_std, series_std_nz, query_mu, query_std, query_std_nz)

Bases: distancematrix.generator.abstract_generator.AbstractBoundStreamingGenerator

Helper class that provides a standard way to create an ABC using inheritance.

append_series(self, values)

Adds more data points to the series sequence (and the query in case of a self-join). Older data points will be dropped if the series would become larger than the foreseen capacity.

Parameters

values – 1D array, the new values to append to the series

Returns

None

append_query(self, values)

Adds more data points to the query sequence. Older data points will be dropped if the query would become larger than the foreseen capacity.

Parameters

values – 1D array, the new values to append to the query

Returns

None

calc_diagonal(self, diag)

calc_column(self, column)

calc_single(self, row, column, dot_prod=None)

Calculates a single point of the distance matrix.

Parameters

• row – index of the subsequence in the query series

• column – index of the subsequence in the data series

• dot_prod – the dotproduct of the subsequences, if provided, this method can run in constant time

Returns

z-normalised distance of the 2 subsequences