Transformer

MultiRocket

Multi RandOm Convolutional KErnel Transform (MultiRocket).

MultiRocket [1] is uses the same set of kernels as MiniRocket on both the raw series and the first order differenced series representation. It uses a different set of dilations and used for each representation. In addition to percentage of positive values (PPV) MultiRocket adds 3 pooling operators: Mean of Positive Values (MPV); Mean of Indices of Positive Values (MIPV); and Longest Stretch of Positive Values (LSPV). This version is for univariate time series only. Use class MultiRocketMultivariate for multivariate input.

This transformer fits one set of paramereters per individual series, and applies the transform with fitted parameter i to the i-th series in transform. Vanilla use requires same number of series in fit and transform.

To fit and transform series at the same time, without an identification of fit/transform instances, wrap this transformer in FitInTransform, from sktime.transformations.compose.

Schnellstart

python

from sktime.transformations.rocket import MultiRocket

estimator = MultiRocket(num_kernels=6250, max_dilations_per_kernel=32, n_features_per_kernel=4, normalise=False, n_jobs=1, random_state=None)

Parameter(6)

num_kernelsint, default = 6,250: number of random convolutional kernels. This should be a multiple of 84. If it is lower than 84, it will be set to 84. If it is higher than 84 and not a multiple of 84, the number of kernels used to transform the data will rounded down to the next positive multiple of 84.
max_dilations_per_kernelint, default = 32: maximum number of dilations per kernel.
n_features_per_kernelint, default = 4: number of features per kernel.
normalisebool, default False
n_jobsint, default=1: The number of jobs to run in parallel for transform. -1 means using all processors.
random_stateNone or int, default = None

Beispiele

>>> from sktime.transformations.rocket import Rocket
>>> from sktime.datasets import load_unit_test
>>> X_train, y_train = load_unit_test (split = "train")
>>> X_test, y_test = load_unit_test (split = "test")
>>> trf = MultiRocket (num_kernels = 512)
>>> trf. fit (X_train) MultiRocket(
... )
>>> X_train = trf. transform (X_train)
>>> X_test = trf. transform (X_test)

Referenzen

[1]

Tan, Chang Wei and Dempster, Angus and Bergmeir, Christoph and

Webb, Geoffrey I, “MultiRocket: Multiple pooling operators and transformations for fast and effective time series classification”,2022, https://link.springer.com/article/10.1007/s10618-022-00844-1 https://arxiv.org/abs/2102.00457