tspy.ml.clustering.kshape module¶

tspy.ml.clustering.kshape.fit(multi_time_series, k_clusters, num_runs, use_eigen=True, init_strategy='plusplus')¶

create a time-series-clustering model using the kshape clustering implementation.

Parameters

multi_time_seriesMultiTimeSeries: the input multi-time-series
k_clustersint or list: if a single int is given, will create a clustering model with the given number of cluster. If a list is given, k-shape will be performed for each k-cluster value between the first number given and the second number given, the output will be the best model based on the min sum of square distances for all centroids
num_runsint: number of runs for clustering
use_eigenbool, optional: if True, uses Eigen-Decomposition for shape extraction, otherwise uses simple averaging (default is True)
init_strategystr, optional: the initialization strategy for the seed centroids in the model. Can be one of “random”, “zero”, “plusplus” (default is “plusplus”)

Returns