BagOfWords#
- class pyts.bag_of_words.BagOfWords(window_size: int | float = 0.5, word_size: int | float = 0.5, n_bins: int = 4, strategy: str = 'normal', numerosity_reduction: bool = True, window_step: int | float = 1, threshold_std: int | float | None = 0.01, norm_mean: bool = True, norm_std: bool = True, overlapping: bool = True, raise_warning: bool = False, alphabet: ArrayLike | None = None)[source]#
Bag-of-words representation for time series.
This algorithm uses a sliding window to extract subsequences from the time series and transforms each subsequence into a word using the Piecewise Aggregate Approximation and the Symbolic Aggregate approXimation algorithms. Thus it transforms each time series into a bag of words.
- Parameters:
- window_sizeint or float (default = 0.5)
Length of the sliding window. If float, it represents a percentage of the size of each time series and must be between 0 and 1.
- word_sizeint or float (default = 0.5)
Length of the words. If float, it represents a percentage of the length of the sliding window and must be between 0. and 1.
- n_binsint (default = 4)
The number of bins to produce. It must be between 2 and
min(n_timestamps, 26).- strategy‘uniform’, ‘quantile’ or ‘normal’ (default = ‘normal’)
Strategy used to define the widths of the bins:
‘uniform’: All bins in each sample have identical widths
‘quantile’: All bins in each sample have the same number of points
‘normal’: Bin edges are quantiles from a standard normal distribution
- numerosity_reductionbool (default = True)
If True, delete sample-wise all but one occurence of back to back identical occurences of the same words.
- window_stepint or float (default = 1)
Step of the sliding window. If float, it represents the percentage of the size of each time series and must be between 0 and 1. The window size will be computed as
ceil(window_step * n_timestamps).- threshold_stdint, float or None (default = 0.01)
Threshold used to determine whether a subsequence is standardized. Subsequences whose standard deviations are lower than this threshold are not standardized. If None, all the subsequences are standardized.
- norm_meanbool (default = True)
If True, center each subseries before scaling.
- norm_stdbool (default = True)
If True, scale each subseries to unit variance.
- overlappingbool (default = True)
If True, time points may belong to two bins when decreasing the size of the subsequence with the Piecewise Aggregate Approximation algorithm. If False, each time point belong to one single bin, but the size of the bins may vary.
- raise_warningbool (default = False)
If True, a warning is raised when the number of bins is smaller for at least one subsequence. In this case, you should consider decreasing the number of bins, using another strategy to compute the bins or removing the corresponding time series.
- alphabetNone or array-like, shape = (n_bins,)
Alphabet to use. If None, the first n_bins letters of the Latin alphabet are used.
References
[1]J. Lin, R. Khade and Y. Li, “Rotation-invariant similarity in time series using bag-of-patterns representation”. Journal of Intelligent Information Systems, 39 (2), 287-315 (2012).
Examples
>>> import numpy as np >>> from pyts.bag_of_words import BagOfWords >>> X = np.arange(12).reshape(2, 6) >>> bow = BagOfWords(window_size=4, word_size=4) >>> bow.transform(X) array(['abcd', 'abcd'], dtype='<U4') >>> bow.set_params(numerosity_reduction=False) BagOfWords(...) >>> bow.transform(X) array(['abcd abcd abcd', 'abcd abcd abcd'], dtype='<U14')
Methods
__init__([window_size, word_size, n_bins, ...])fit(X[, y])Pass.
fit_transform(X[, y])Fit to data, then transform it.
get_metadata_routing()Get metadata routing of this object.
get_params([deep])Get parameters for this estimator.
set_output(*[, transform])Set output container.
set_params(**params)Set the parameters of this estimator.
transform(X)Transform each time series into a bag of words.
- __init__(window_size: int | float = 0.5, word_size: int | float = 0.5, n_bins: int = 4, strategy: str = 'normal', numerosity_reduction: bool = True, window_step: int | float = 1, threshold_std: int | float | None = 0.01, norm_mean: bool = True, norm_std: bool = True, overlapping: bool = True, raise_warning: bool = False, alphabet: ArrayLike | None = None) None[source]#
