distancematrix.calculator

Module Contents

Classes

AbstractCalculator	Base class for calculators. A calculator is repsonsible for managing
AnytimeCalculator	Calculator that allows approximate calculations in a fraction of the time, but does not support
StreamingCalculator	Calculator that allows streaming data, but does not support anytime calculations.

class distancematrix.calculator.AbstractCalculator(self_join, m, num_series_subseq, num_query_subseq, n_dim, trivial_match_buffer=None)

Bases: abc.ABC

Base class for calculators. A calculator is repsonsible for managing consumers and generators for a distance matrix calculation. It provides a single point of interaction for the user.

In order to do useful work, generators and consumers need to be added. Generators will use the input query and series to form a distance matrix. Consumers process these values in a way that is useful.

add_consumer(self, generator_ids, consumer)

Adds a consumer that uses the distances calculated by the provided generators.

Parameters

• generator_ids – list containing ids of the generators

• consumer – the consumer to add

Returns

the bound consumer

abstract add_generator(self, input_dim, generator)

Adds a generator that will use the data from the specified channel (from series/query).

Parameters

• input_dim – index of the data channel

• generator – the generator to add

Returns

the bound generator

calculate_columns(self, start=None, upto=1.0, print_progress=False)

Calculates columns of the distance matrix. The calculator keeps track of the rightmost column that was already calculated and will use it as starting position unless the start position is provided.

Note that the generators are optimised for calculating consecutive columns from left to right.

Parameters

start – int for absolute position, float for relative position. The first column to calculate.

If None, continues from the rightmost column that was calculated so far. :param upto: int for absolute position, float for relative position. The last column (exclusive) to calculate. :param print_progress: print progress to console? :return: None

property num_dist_matrix_values(self)

property generators(self)

property consumers(self)

class distancematrix.calculator.AnytimeCalculator(m, series, query=None, trivial_match_buffer=None)

Bases: distancematrix.calculator.AbstractCalculator

Calculator that allows approximate calculations in a fraction of the time, but does not support data streaming.

A calculator is repsonsible for managing consumers and generators for a distance matrix calculation. It provides a single point of interaction for the user.

add_generator(self, input_dim, generator)

Adds a generator that will use the data from the specified channel (from series/query).

Parameters

• input_dim – index of the data channel

• generator – the generator to add

Returns

the bound generator

calculate_diagonals(self, partial=1.0, print_progress=False)

Calculates diagonals of the distance matrix. The advantage of calculating diagonals is that values are spread over the entire distance matrix, which can provide a quick approximation for any consumer.

Parameters

partial – int for a number of values, float for relative number of values. The number of distance

matrix values that should be calculated (including the counts of previous diagonals calulated). :param print_progress: print progress to the console :return: None

class distancematrix.calculator.StreamingCalculator(m, series_window, query_window=None, n_dim=1, trivial_match_buffer=None)

Bases: distancematrix.calculator.AbstractCalculator

Calculator that allows streaming data, but does not support anytime calculations.

A calculator is repsonsible for managing consumers and generators for a distance matrix calculation. It provides a single point of interaction for the user.

add_generator(self, input_dim, generator)

Adds a generator that will use the data from the specified channel (from series/query).

Parameters

• input_dim – index of the data channel

• generator – the generator to add

Returns

the bound generator

append_series(self, values)

Add more data points to series.

As a side effect, the last calculated column index is shifted along with the data.

Parameters

values – 1D array for one data point on each channel, or 2D array of shape (num_dim, num_points)

Returns

None

append_query(self, values)

Add more data points to query. Cannot be used if performing a self join.

Note that appending query data does not adjust the last column calculated index.

Parameters

values – 1D array for one data point on each channel, or 2D array of shape (num_dim, num_points)

Returns

None