distancematrix.consumer.matrix_profile_lr

Module Contents

Classes

MatrixProfileLR	Consumer of distance matrix information to form the left and right matrix profile and their corresponding
ShiftingMatrixProfileLR	Extension of MatrixProfileLR which supports streaming.
MatrixProfileLRReservoir	Consumer of distance matrix information to form the left and right matrix profile and their corresponding

class distancematrix.consumer.matrix_profile_lr.MatrixProfileLR

Bases: distancematrix.consumer.abstract_consumer.AbstractConsumer

Consumer of distance matrix information to form the left and right matrix profile and their corresponding indices. The left matrix profile and index is a one dimensional series where each each value at index i contains the distance and index of the best match whose index is equal to or less than i. The right matrix profile and index contain the best match whose index is larger than i.

initialise(self, dims, query_subseq, series_subseq)

Initialise this consumer.

Parameters

• dims – the number of dimensions (data channels) this consumer will receive

• query_subseq – the number of query subsequences (rows in the distance matrix)

• series_subseq – the number of series subsequences (column in the distance matrix)

Returns

None

process_diagonal(self, diag, values)

Method called when a diagonal of the distance matrix is calculated.

The number of values on the diagonal might be less than the diagonal of the full matrix profile, this can occur when not enough data is available yet to calculate the entire distance matrix (typically for streaming when not enough data is available to fill the entire foreseen space).

Parameters

• diagonal_index – index of the diagonal in range ]-num_query_subseq, num_series_subseq[, the main diagonal has index 0

• values – array of shape (num_dimensions, num_values_on_diagonal) containing the distances

Returns

None

process_column(self, column_index, values)

Method called when a column of the distance matrix is calculated.

The number of values on the column might be less than the column of the full matrix profile, this can occur when not enough data is available yet to calculate the entire distance matrix (typically for streaming when not enough data is available to fill the entire foreseen space).

Parameters

• column_index – index of the column, in range [0, series_subseq[

• values – array of shape (num_dimensions, num_values_on_column) containing the distances

Returns

None

property matrix_profile_right(self)

property matrix_profile_left(self)

property profile_index_right(self)

property profile_index_left(self)

matrix_profile(self)

Creates the matrix profile based on the left and right matrix profile.

Returns

1D array

profile_index(self)

Creates the matrix profile index based on the left and right indices.

Returns

1D array

class distancematrix.consumer.matrix_profile_lr.ShiftingMatrixProfileLR(rb_scale_factor=2.0)

Bases: distancematrix.consumer.matrix_profile_lr.MatrixProfileLR, distancematrix.consumer.abstract_consumer.AbstractStreamingConsumer

Extension of MatrixProfileLR which supports streaming.

The profile indices tracked in this consumer refer to positions in the complete query series. As an example, if the original query consisted of 10 sequences, but has since shifted by 5 sequences, the profile indices can contain any value in [0..15], or -1 if no matrix profile value exists. These indices can be converted to indices local to the current window by subtracting the query_shift, keep in mind that some indices of the left matrix profile can refer to positions outside the current window.

The concept of left and right matrix profile is only useful when both query and series shift at the same time (distances are calculated over a self-join). Even if this is not the case, the values in this consumer are correct: the left matrix profile stores any values on or above the (unshifted) main diagonal, the right matrix profile stores any values below the (unshifted) main diagonal. (Since the diagonal shifts away when only the series is shifted, eventually only the left matrix profile will be used.)

initialise(self, dims, query_subseq, series_subseq)

Initialise this consumer.

Parameters

• dims – the number of dimensions (data channels) this consumer will receive

• query_subseq – the number of query subsequences (rows in the distance matrix)

• series_subseq – the number of series subsequences (column in the distance matrix)

Returns

None

process_diagonal(self, diag, values)

Method called when a diagonal of the distance matrix is calculated.

The number of values on the diagonal might be less than the diagonal of the full matrix profile, this can occur when not enough data is available yet to calculate the entire distance matrix (typically for streaming when not enough data is available to fill the entire foreseen space).

Parameters

• diagonal_index – index of the diagonal in range ]-num_query_subseq, num_series_subseq[, the main diagonal has index 0

• values – array of shape (num_dimensions, num_values_on_diagonal) containing the distances

Returns

None

process_column(self, column_index, values)

Method called when a column of the distance matrix is calculated.

The number of values on the column might be less than the column of the full matrix profile, this can occur when not enough data is available yet to calculate the entire distance matrix (typically for streaming when not enough data is available to fill the entire foreseen space).

Parameters

• column_index – index of the column, in range [0, series_subseq[

• values – array of shape (num_dimensions, num_values_on_column) containing the distances

Returns

None

shift_query(self, amount)

Inform the consumer that the distance matrix has shifted in the query direction.

Parameters

amount – amount of subsequences that were shifted

Returns

None

shift_series(self, amount)

Inform the consumer that the distance matrix has shifted in the series direction.

Parameters

amount – amount of subsequences that were shifted

Returns

None

property matrix_profile_right(self)

property matrix_profile_left(self)

property profile_index_right(self)

property profile_index_left(self)

class distancematrix.consumer.matrix_profile_lr.MatrixProfileLRReservoir(random_seed=None)

Bases: distancematrix.consumer.abstract_consumer.AbstractConsumer

Consumer of distance matrix information to form the left and right matrix profile and their corresponding indices. The left matrix profile and index is a one dimensional series where each each value at index i contains the distance and index of the best match whose index is less than i. The right matrix profile and index contain the best match whose index is equal to or larger than i.

This consumer differs from the regular MatrixProfileLR consumer in that it uses reservoir sampling to determine the matrix profile indices. This means that if multiple values in the distance matrix column have the exact same distance, a random index will be stored in the matrix profile index.

initialise(self, dims, query_subseq, series_subseq)

Initialise this consumer.

Parameters

• dims – the number of dimensions (data channels) this consumer will receive

• query_subseq – the number of query subsequences (rows in the distance matrix)

• series_subseq – the number of series subsequences (column in the distance matrix)

Returns

None

process_diagonal(self, diag, values)

Method called when a diagonal of the distance matrix is calculated.

The number of values on the diagonal might be less than the diagonal of the full matrix profile, this can occur when not enough data is available yet to calculate the entire distance matrix (typically for streaming when not enough data is available to fill the entire foreseen space).

Parameters

• diagonal_index – index of the diagonal in range ]-num_query_subseq, num_series_subseq[, the main diagonal has index 0

• values – array of shape (num_dimensions, num_values_on_diagonal) containing the distances

Returns

None

process_column(self, column_index, values)

Method called when a column of the distance matrix is calculated.

The number of values on the column might be less than the column of the full matrix profile, this can occur when not enough data is available yet to calculate the entire distance matrix (typically for streaming when not enough data is available to fill the entire foreseen space).

Parameters

• column_index – index of the column, in range [0, series_subseq[

• values – array of shape (num_dimensions, num_values_on_column) containing the distances

Returns

None

matrix_profile(self)

Creates the matrix profile based on the left and right matrix profile.

Returns

1D array

profile_index(self)

Creates the matrix profile index based on the left and right indices.

Returns

1D array