pyredraw.ts

Module Contents

Functions

moving_block_selection(data_array, block_window_length)

Select sample of blocks from input time series array via moving block bootstrap.

moving_block_bootstrap(data_array, ...[, seed])

Create replicates of time series array via moving block bootstrap.
pyredraw.ts.moving_block_selection(data_array, block_window_length)[source]
Select sample of blocks from input time series array via moving block bootstrap.

Result is a single replicate of the input array.

Parameters

  • data_array (numpy ndarray (1D)) – Time series data on which user would like perform moving block bootstrap.

  • block_window_length (int) – Number of data points in a block bootstrap sample. Length of block sample. If input is float, will be converted to integer.

Returns

Moving block boostrap result from input data_array. Result is same size as input data given by data_array and is a single boostrap replicate of the input data_array.

Return type

numpy ndarray (1D)

Examples

>>> data_array = [1,2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12]
>>> block_window_length = 3
>>> moving_block_selection(data_array, block_window_length)

References

Bergmeir C., Hyndman R.J., Benitez J.M.: Bagging exponential smoothing methods using STL decomposition and Box-cox Transformation. International Journal of Forecasting, 32(2):303-312, (2016)

pyredraw.ts.moving_block_bootstrap(data_array, block_window_length, number_replicates, seed=1027)[source]

Create replicates of time series array via moving block bootstrap.

Parameters

  • data_array (list, pandas Series, numpy ndarray (1D)) – Time series data on which user would like perform moving block bootstrap.

  • block_window_length (int or float) – Number of data points in a block bootstrap sample. Length of block sample. If input is float, will be converted to integer.

  • number_replicates (int or float) – Number of bootstrap replicates of input data to create.

  • seed (int) – Value of random seed for random number generator for reproducibility. Default value of random seed is 1027 which will be used if no value is provided.

Returns

Moving block boostrap replicates from input data_array. Resulting array has number of columns equal to length of input data and number of rows is equal to number of replicates. Each row is a single bootstrap replicate.

Return type

numpy ndarray (2D)

Examples

>>> data_array = [1,2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12]
>>> block_window_length = 3
>>> number_replicates = 10
>>> moving_block_bootstrap(data_array, block_window_length, number_replicates)