trajminer.clustering.KMedoids

class trajminer.clustering.KMedoids(n_clusters, init=None, seed=None, max_iter=300, measure='precomputed', n_jobs=1)

K-Medoids Clustering.

n_clustersint

The number of clusters to group trajectories into.

init‘park’ or array-like (default=None)

The indices of the trajectories representing the initial cluster medoids. If ‘park’, the medoids will be initialized using the approach introduced in [Park et al., 2009] (see references). If None, the initial medoids will be chosen randomly.

seedint (default=None)

The random seed to be used for centroid initialization. If None, the default seed of NumPy will be used.

max_iterint (default=300)

The maximum number of iterations to run the algorithm, in case it has not yet converged.

measureSimilarityMeasure object or str (default=’precomputed’)

The similarity measure to use for computing similarities (see trajminer.similarity) or the string ‘precomputed’.

n_jobsint (default=1)

The number of parallel jobs.

Park, H. S., & Jun, C. H. (2009). A simple and fast algorithm for K-medoids clustering. Expert systems with applications, 36(2), 3336-3341.

__init__(n_clusters, init=None, seed=None, max_iter=300, measure='precomputed', n_jobs=1)

Initialize self. See help(type(self)) for accurate signature.

fit_predict(X)

Fits and returns the predictions for the given test data.

Xarray-like, shape (n_samples, max_length, n_features)

Input data. If measure == ‘precomputed’, then X is a distance matrix with shape (n_samples, n_samples).

predictionsarray-like, shape (n_samples)

Assigned cluster for each input sample.