WEASEL+MUSE

WEASELMUSE stand for Word ExtrAction for time SEries cLassification plus Multivariate Unsupervised Symbols and dErivatives. This example shows how the WEASELMUSE algorithm transforms multivariate time series of real numbers into a sequence of frequencies of words, and illustrates the features obtained for two time series. It is implemented as pyts.multivariate.transformation.WEASELMUSE.

# Author: Johann Faouzi <johann.faouzi@gmail.com>
# License: BSD-3-Clause

import matplotlib.pyplot as plt
import numpy as np
from sklearn.preprocessing import LabelEncoder

from pyts.datasets import load_basic_motions
from pyts.multivariate.transformation import WEASELMUSE

# Toy dataset
X_train, _, y_train, _ = load_basic_motions(return_X_y=True)
y_train = LabelEncoder().fit_transform(y_train)

# WEASEL+MUSE transformation
transformer = WEASELMUSE(
    word_size=2,
    n_bins=2,
    window_sizes=[12, 36],
    chi2_threshold=15,
    sparse=False,
)
X_weasel = transformer.fit_transform(X_train, y_train)

# Visualize the transformation for the first time series
plt.figure(figsize=(6, 4))
vocabulary_length = len(transformer.vocabulary_)
width = 0.3
plt.bar(
    np.arange(vocabulary_length) - width / 2,
    X_weasel[y_train == 0][0],
    width=width,
    label='First time series in class 0',
)
plt.bar(
    np.arange(vocabulary_length) + width / 2,
    X_weasel[y_train == 1][0],
    width=width,
    label='First time series in class 1',
)
plt.xticks(
    np.arange(vocabulary_length),
    np.vectorize(transformer.vocabulary_.get)(np.arange(X_weasel[0].size)),
    fontsize=12,
    rotation=60,
    ha='right',
)
y_max = np.max(np.concatenate([X_weasel[y_train == 0][0], X_weasel[y_train == 1][0]]))
plt.yticks(np.arange(y_max + 1), fontsize=12)
plt.xlabel("Words", fontsize=14)
plt.ylabel("Frequencies", fontsize=14)
plt.title("WEASEL+MUSE transformation", fontsize=16)
plt.legend(loc='best', fontsize=10)

plt.subplots_adjust(bottom=0.27)
plt.tight_layout()
plt.show()