sparse_affinity_propagation.py 文件源码

def fit(self, X, **kwargs):
        """Apply affinity propagation clustering.

        Create affinity matrix from negative euclidean distances if required.

        Parameters
        ----------
        X: array-like or sparse matrix,
                shape (n_samples, n_features) or (n_samples, n_samples)
            Data matrix or, if affinity is ``precomputed``, matrix of
            similarities / affinities.
        """
        if not issparse(X):
            return super(AffinityPropagation, self).fit(X, **kwargs)

        # Since X is sparse, this converts it in a coo_matrix if required
        X = check_array(X, accept_sparse='coo')
        if self.affinity == "precomputed":
            self.affinity_matrix_ = X
        elif self.affinity == "euclidean":
            self.affinity_matrix_ = coo_matrix(
                -euclidean_distances(X, squared=True))
        else:
            raise ValueError("Affinity must be 'precomputed' or "
                             "'euclidean'. Got %s instead"
                             % str(self.affinity))

        self.cluster_centers_indices_, self.labels_, self.n_iter_ = \
            sparse_ap(
                self.affinity_matrix_, self.preference, max_iter=self.max_iter,
                convergence_iter=self.convergence_iter, damping=self.damping,
                copy=self.copy, verbose=self.verbose, return_n_iter=True,
                convergence_percentage=self.convergence_percentage)

        if self.affinity != "precomputed":
            self.cluster_centers_ = X.data[self.cluster_centers_indices_].copy()

        return self