ó ‡ˆ\c@ s>ddlmZddlZddlZddlmZddlmZddl m Z ddl m Z m Z mZddlmZdd lmZdd lmZmZdd lmZdd lmZmZd dlmZd„Zd„Zdefd„ƒYZ defd„ƒYZ!dee e fd„ƒYZ"dS(iÿÿÿÿ(tdivisionN(tsparse(tsqrti(teuclidean_distances(tTransformerMixint ClusterMixint BaseEstimator(txrange(t check_array(t row_normstsafe_sparse_dot(tcheck_is_fitted(tNotFittedErrortConvergenceWarningi(tAgglomerativeClusteringc c s•|jd}|j}|j}|j}xft|ƒD]X}tj|jdƒ}||||d}}|||!} |||!|| <|Vq5WdS(s£This little hack returns a densified row when iterating over a sparse matrix, instead of constructing a sparse matrix for every row that is expensive. iiN(tshapetindicestdatatindptrRtnptzeros( tXt n_samplest X_indicestX_datatX_indptrtitrowtstartptrtendptrtnonzero_indices((s4lib/python2.7/site-packages/sklearn/cluster/birch.pyt_iterate_sparse_Xs     cC s«tƒ}tƒ}t||d|jd|jƒ}t||d|jd|jƒ}||_||_|jrØ|jdk r||j_n|j|_||_||_|j|_|jdk rØ||j_qØnt|j d|j dt ƒ}|j d}t j|jƒ||fƒ} || f\} } | | k} x^t|jƒD]M\} }| | rƒ|j|ƒ|j|ƒqP|j|ƒ|j|ƒqPW||fS(s¾The node has to be split if there is no place for a new subcluster in the node. 1. Two empty nodes and two empty subclusters are initialized. 2. The pair of distant subclusters are found. 3. The properties of the empty subclusters and nodes are updated according to the nearest distance between the subclusters to the pair of distant subclusters. 4. The two nodes are set as children to the two subclusters. tis_leaft n_featurestY_norm_squaredtsquarediN(t _CFSubclustert_CFNodeR R!tchild_t prev_leaf_tNonet next_leaf_Rt centroids_t squared_norm_tTrueRRt unravel_indextargmaxt enumeratet subclusters_tappend_subclustertupdate(tnodet thresholdtbranching_factortnew_subcluster1tnew_subcluster2t new_node1t new_node2tdistt n_clusterst farthest_idxt node1_distt node2_distt node1_closertidxt subcluster((s4lib/python2.7/site-packages/sklearn/cluster/birch.pyt _split_node(sB                R%cB s2eZdZd„Zd„Zd„Zd„ZRS(sEach node in a CFTree is called a CFNode. The CFNode can have a maximum of branching_factor number of CFSubclusters. Parameters ---------- threshold : float Threshold needed for a new subcluster to enter a CFSubcluster. branching_factor : int Maximum number of CF subclusters in each node. is_leaf : bool We need to know if the CFNode is a leaf or not, in order to retrieve the final subclusters. n_features : int The number of features. Attributes ---------- subclusters_ : array-like list of subclusters for a particular CFNode. prev_leaf_ : _CFNode prev_leaf. Useful only if is_leaf is True. next_leaf_ : _CFNode next_leaf. Useful only if is_leaf is True. the final subclusters. init_centroids_ : ndarray, shape (branching_factor + 1, n_features) manipulate ``init_centroids_`` throughout rather than centroids_ since the centroids are just a view of the ``init_centroids_`` . init_sq_norm_ : ndarray, shape (branching_factor + 1,) manipulate init_sq_norm_ throughout. similar to ``init_centroids_``. centroids_ : ndarray view of ``init_centroids_``. squared_norm_ : ndarray view of ``init_sq_norm_``. cC s~||_||_||_||_g|_tj|d|fƒ|_tj|dƒ|_g|_ d|_ d|_ dS(Ni( R4R5R R!R0RRtinit_centroids_t init_sq_norm_R+R(R'R)(tselfR4R5R R!((s4lib/python2.7/site-packages/sklearn/cluster/birch.pyt__init__‰s       cC s}t|jƒ}|jj|ƒ|j|j|<|j|j|<|jd|d…dd…f|_|j|d |_dS(Ni( tlenR0tappendt centroid_RCtsq_norm_RDR*R+(RERAR((s4lib/python2.7/site-packages/sklearn/cluster/birch.pyR1˜s &cC sP|jj|ƒ}||j|<|j|j|<|j|j|<|j|ƒdS(sZRemove a subcluster from a node and update it with the split subclusters. N(R0tindexRIRCRJRDR1(RERAR6R7tind((s4lib/python2.7/site-packages/sklearn/cluster/birch.pytupdate_split_subclusters¤s  c C s«|js|j|ƒtS|j}|j}tj|j|jƒ}|d9}||j 7}tj |ƒ}|j|}|j dk r.|j j |ƒ}|sÝ|j|ƒ|j|j|j|<|j|j|j|`. Parameters ---------- threshold : float, default 0.5 The radius of the subcluster obtained by merging a new sample and the closest subcluster should be lesser than the threshold. Otherwise a new subcluster is started. Setting this value to be very low promotes splitting and vice-versa. branching_factor : int, default 50 Maximum number of CF subclusters in each node. If a new samples enters such that the number of subclusters exceed the branching_factor then that node is split into two nodes with the subclusters redistributed in each. The parent subcluster of that node is removed and two new subclusters are added as parents of the 2 split nodes. n_clusters : int, instance of sklearn.cluster model, default 3 Number of clusters after the final clustering step, which treats the subclusters from the leaves as new samples. - `None` : the final clustering step is not performed and the subclusters are returned as they are. - `sklearn.cluster` Estimator : If a model is provided, the model is fit treating the subclusters as new samples and the initial data is mapped to the label of the closest subcluster. - `int` : the model fit is :class:`AgglomerativeClustering` with `n_clusters` set to be equal to the int. compute_labels : bool, default True Whether or not to compute labels for each fit. copy : bool, default True Whether or not to make a copy of the given data. If set to False, the initial data will be overwritten. Attributes ---------- root_ : _CFNode Root of the CFTree. dummy_leaf_ : _CFNode Start pointer to all the leaves. subcluster_centers_ : ndarray, Centroids of all subclusters read directly from the leaves. subcluster_labels_ : ndarray, Labels assigned to the centroids of the subclusters after they are clustered globally. labels_ : ndarray, shape (n_samples,) Array of labels assigned to the input data. if partial_fit is used instead of fit, they are assigned to the last batch of data. Examples -------- >>> from sklearn.cluster import Birch >>> X = [[0, 1], [0.3, 1], [-0.3, 1], [0, -1], [0.3, -1], [-0.3, -1]] >>> brc = Birch(branching_factor=50, n_clusters=None, threshold=0.5, ... compute_labels=True) >>> brc.fit(X) # doctest: +NORMALIZE_WHITESPACE Birch(branching_factor=50, compute_labels=True, copy=True, n_clusters=None, threshold=0.5) >>> brc.predict(X) array([0, 0, 0, 1, 1, 1]) References ---------- * Tian Zhang, Raghu Ramakrishnan, Maron Livny BIRCH: An efficient data clustering method for large databases. http://www.cs.sfu.ca/CourseCentral/459/han/papers/zhang96.pdf * Roberto Perdisci JBirch - Java implementation of BIRCH clustering algorithm https://code.google.com/archive/p/jbirch Notes ----- The tree data structure consists of nodes with each node consisting of a number of subclusters. The maximum number of subclusters in a node is determined by the branching factor. Each subcluster maintains a linear sum, squared sum and the number of samples in that subcluster. In addition, each subcluster can also have a node as its child, if the subcluster is not a member of a leaf node. For a new point entering the root, it is merged with the subcluster closest to it and the linear sum, squared sum and the number of samples of that subcluster are updated. This is done recursively till the properties of the leaf node are updated. gà?i2icC s1||_||_||_||_||_dS(N(R4R5R;tcompute_labelstcopy(RER4R5R;RjRk((s4lib/python2.7/site-packages/sklearn/cluster/birch.pyRF¬s     cC s tt|_|_|j|ƒS(sÑ Build a CF Tree for the input data. Parameters ---------- X : {array-like, sparse matrix}, shape (n_samples, n_features) Input data. y : Ignored (R,RNtfit_t partial_fit_t_fit(RERty((s4lib/python2.7/site-packages/sklearn/cluster/birch.pytfit´s cC sôt|ddd|jƒ}|j}|j}|dkrHtdƒ‚n|j\}}t|dƒ}t|ddƒ}t|dƒs”|rñ| rñt||d t d |ƒ|_ t||d t d |ƒ|_ |j |j _ |j |j _ ntj|ƒs t}nt}x||ƒD]} td | ƒ} |j j| ƒ} | rt|j ||ƒ\} } |` t||d td |ƒ|_ |j j| ƒ|j j| ƒqqWtjg|jƒD]}|j^qƒ}||_|j|ƒ|S( Nt accept_sparsetcsrRkis,Branching_factor should be greater than one.Rmtroot_RlR R!R^(RRkR4R5t ValueErrorRtgetattrR(R%R,Rst dummy_leaf_R)R'RtissparsetiterRR$RQRBRNR1Rt concatenatet _get_leavesR*tsubcluster_centers_t_global_clustering(RERR4R5RR!t partial_fitthas_roott iter_functsampleRAtsplitR6R7tleaft centroids((s4lib/python2.7/site-packages/sklearn/cluster/birch.pyRnÃsF        %  cC s?|jj}g}x&|dk r:|j|ƒ|j}qW|S(s– Retrieve the leaves of the CF Node. Returns ------- leaves : array-like List of the leaf nodes. N(RvR)R(RH(REtleaf_ptrtleaves((s4lib/python2.7/site-packages/sklearn/cluster/birch.pyRzös   cC sKtt|_|_|dkr-|jƒ|S|j|ƒ|j|ƒSdS(s; Online learning. Prevents rebuilding of CFTree from scratch. Parameters ---------- X : {array-like, sparse matrix}, shape (n_samples, n_features), None Input data. If X is not provided, only the global clustering step is done. y : Ignored N(R,RNRmRlR(R|t _check_fitRn(RERRo((s4lib/python2.7/site-packages/sklearn/cluster/birch.pyR}s    cC sot|dƒ}t|dƒ}|p'|s9tdƒ‚n|rk|jd|jjdkrktdƒ‚ndS(NR{Rms#Fit training data before predictingisETraining data and predicted data do not have same number of features.(thasattrR RR{Rt(RERt is_fittedthas_partial_fit((s4lib/python2.7/site-packages/sklearn/cluster/birch.pyR†s #cC set|ddƒ}|j|ƒt||jjƒ}|d9}||j7}|jtj|ddƒS(sf Predict data using the ``centroids_`` of subclusters. Avoid computation of the row norms of X. Parameters ---------- X : {array-like, sparse matrix}, shape (n_samples, n_features) Input data. Returns ------- labels : ndarray, shape(n_samples) Labelled data. RqRriþÿÿÿtaxisi( RR†R R{tTt_subcluster_normstsubcluster_labels_RRP(RERtreduced_distance((s4lib/python2.7/site-packages/sklearn/cluster/birch.pytpredict+s    cC st|dƒt||jƒS(s¼ Transform X into subcluster centroids dimension. Each dimension represents the distance from the sample point to each cluster centroid. Parameters ---------- X : {array-like, sparse matrix}, shape (n_samples, n_features) Input data. Returns ------- X_trans : {array-like, sparse matrix}, shape (n_samples, n_clusters) Transformed data. R{(R RR{(RER((s4lib/python2.7/site-packages/sklearn/cluster/birch.pyt transformBs cC s?|j}|j}|dk o$|j}t}t|tƒrotd|jƒ}t|ƒ|jkršt }qšn+|dk ršt |dƒ ršt dƒ‚nt |jdt ƒ|_ |dksÄ|r tjt|ƒƒ|_|r tjdt|ƒ|jftƒq n|j|jƒ|_|r;|j|ƒ|_ndS(sN Global clustering for the subclusters obtained after fitting R;t fit_predicts:n_clusters should be an instance of ClusterMixin or an intR#sTNumber of subclusters found (%d) by Birch is less than (%d). Decrease the threshold.N(R;R{R(RjRNt isinstancetintRRGR,R‡RtR RŒRtarangeRtwarningstwarnR R‘Rtlabels_(RERt clustererRƒRjtnot_enough_centroids((s4lib/python2.7/site-packages/sklearn/cluster/birch.pyR|Vs0      N(RXRYRZR,RFR(RpRnRzR}R†RRR|(((s4lib/python2.7/site-packages/sklearn/cluster/birch.pyRiDsf   3    (#t __future__RR•tnumpyRtscipyRtmathRtmetrics.pairwiseRtbaseRRRtexternals.six.movesRtutilsRt utils.extmathR R tutils.validationR t exceptionsR R t hierarchicalRRRBtobjectR%R$Ri(((s4lib/python2.7/site-packages/sklearn/cluster/birch.pyts"    2™Q