e52e9c8d7536b6315da655164951060642a52707,sklearn/cluster/dbscan_.py,DBSCAN,fit,#DBSCAN#Any#Any#Any#,284

Before Change

            X_mask = X.data <= self.eps
            masked_indices = X.indices.astype(np.intp, copy=False)[X_mask]
            masked_indptr = np.concatenate(([0], np.cumsum(X_mask)))
            masked_indptr = masked_indptr[X.indptr[1:-1]]

            // split into rows
            neighborhoods[:] = np.split(masked_indices, masked_indptr)
        else:
            neighbors_model = NearestNeighbors(
                radius=self.eps, algorithm=self.algorithm,
                leaf_size=self.leaf_size, metric=self.metric,

After Change

        // Calculate neighborhood for all samples. This leaves the original
        // point in, which needs to be considered later (i.e. point i is in the
        // neighborhood of point i. While True, its useless information)
        if self.metric == "precomputed" and sparse.issparse(X):
            // set the diagonal to explicit values, as a point is its own
            // neighbor
            with warnings.catch_warnings():
                warnings.simplefilter("ignore", sparse.SparseEfficiencyWarning)
                X.setdiag(X.diagonal())  // XXX: modifies X"s internals in-place

        neighbors_model = NearestNeighbors(
            radius=self.eps, algorithm=self.algorithm,
            leaf_size=self.leaf_size, metric=self.metric,
            metric_params=self.metric_params, p=self.p, n_jobs=self.n_jobs)

Instances