a )a"@sddlZddlZddlZddlZddlmZmZddlm Z ddl m Z m Z m Z ddlZddlmZmZddlmZddlmZdd lmZdd lmZmZmZdd lmZmZm Z dd l!m"Z"dd l#m$Z$ddl%m&Z&ddl'm(Z)ddl*m+Z+ddl,m-Z-ddl.m/Z0d.ddZ1ddZ2ddZ3d/ddZ4ddZ5ddZ6d0d d!Z7d1d"d#Z8d2d$d%Z9d&d'Z:d3d(d)Z;Gd*d+d+eZGd,d-d-eZdS)4N) getFPTypeget_patch_message)support_usm_ndarray)daal_check_versionsklearn_check_versionPatchingConditionsChain)DecisionTreeClassifierDecisionTreeRegressor)Tree)RandomForestClassifier)RandomForestRegressor)check_random_state check_array deprecated)check_is_fittedcheck_consistent_length _num_samples)_daal_num_features)clone)DataConversionWarning) __version__) LooseVersion)ceil)sparseFcCs|dur |St|trz|dkr:|r6tdtt|S|S|dkrVtdtt|S|dkrrtdtt|Stdt|tj tj fr|S|dkrtdt||SdS)Nautosqrtlog2zSInvalid value for max_features. Allowed string values are "auto", "sqrt" or "log2".r) isinstancestrmaxintnprr ValueErrornumbersIntegralZinteger) max_featuresZ n_featuresis_classificationr*?lib/python3.9/site-packages/daal4py/sklearn/ensemble/_forest.py_to_absolute_max_features1s( r,cCs|dur dSt|tjrNd|kr,|ksBnd}t|||t||St|tjrtdrdt|krzdksnd}t||n,dt|krdksnd}t||t|Sd}t|t |dS) N?rz7`max_samples` must be in range 1 to {} but got value {}1.0rz:`max_samples` must be in range (0.0, 1.0] but got value {}z6`max_samples` must be in range (0, 1) but got value {}z7`max_samples` should be int or float, but got type '{}') r r&r'r%formatfloatZRealr TypeErrortype) n_samples max_samplesmsgr*r*r+_get_n_samples_bootstrapJs$   r6cCst|jtjr(d|jksNtd|jn&d|jkr>dksNntd|jt|jtjrvd|jkstd|jn&d|jkrdksntd|jd |jkrdksntd |jdurt d t |jdkrtd |j dkrtd |j durr?r@rA_cached_estimators_computemodel daal_model_ZoutOfBagErrorAccuracy oob_score_ZoutOfBagErrorDecisionFunctionZoob_decision_function_squeeze)rCXy sample_weightZexpanded_class_weightrers_seed_X_fptypemax_stream_count daal_engineZfeatures_per_node_Zn_samples_bootstrap_ dfc_algorithmZdfc_trainingResultrur*r*r+_daal_fit_classifiers       #rcCs~t|}tjt|j|dd}|jd|jkrLtd|jdd|jd|||j }|j }t |j |jt jddS) NZcomputeClassLabelsrRrKZresultsToEvaluaterX has 3 features, but RandomForestClassifier is expecting  features as inputZunsafe)Zcasting)rrm)decision_forest_classification_predictionr#rerfrgr%rtrv predictionr$Ztakeclasses_ravelZastypeZint64rCryr~rZdfc_predictionResultpredr*r*r+_daal_predict_classifiers"rcCs6t|}tjt|j|dd}|||j}|j}|S)NZcomputeClassProbabilitiesr)rrmrr#rertrvZ probabilitiesrr*r*r+_daal_predict_probasrc Cst|rtdt||dur,t||}td}||jrFtdpL|j df|j dudf|j dkd|j d f|j d kd |j d ft| d fg}|rZt dr|j |ddt|tjtjgd}t|}t|}|jdkr |jddkr tjdtddt|||jdkr.t|d}|jd|_||jdkd|jdfg}||rt||||d|j|_|j d|_ |j!d|_!|St"t#|j$|||dS)N3sparse multilabel-indicator for y is not supported.z+sklearn.ensemble.RandomForestClassifier.fitPiCOOB score is only supported starting from 2021.5 version of oneDAL.FWarm start is not supported.gini'zA' criterion is not supported. Only 'gini' criterion is supported.rNon-zero 'ccp_alpha' () is not supported.+X is sparse. Sparse input is not supported.r.TresetrGrrA column-vector y was passed when a 1d array was expected. Please change the shape of y to (n_samples,), for example using ravel(). stacklevelrcrNumber of outputs ( ) is not 1.r{r)%spissparser%rDcheck_sample_weightrand_conditionsrpr warm_start criterion ccp_alphar_check_feature_namesrr$float32float64asarray atleast_1dndimrfr;r<rrreshape n_outputs_ write_logr _estimators_ estimators_rersuperr fitrCryrzr{_patching_status _dal_readyr*r*r+_fit_classifier%sd               rcCs|jd|_tds|j|_t|j}|js:|jr:tdt |}| dt dj }d}|j|krttj||d}ntj||d}t|j|jddd } t|jd|jd } |dur|g}tjt |d t|j|jd ur| nd t| t|jdurdn|jt|jtjr|jntt|j|jd|t|jdur:dn|jd|jrNdnddt |jt|j!tjrp|j!ntt|j!|jd|j"|j#|j$durdn|j$|j%|j&d} d|_'| (|||} | j)} | |_*|jr| j+dd|_,| j-j.dd|_/|j/jddkr|j/j.dd|_/|S)Nrr.rNrrHrIrJFrLrMrOTr-rrPz5computeOutOfBagErrorR2|computeOutOfBagErrorPredictionrQ)rKrSrTrUrVrWrXrYrZr[r\r]r^r_r`rarbr@rArdrc)0rfrgrrhrrir^rpr%rrjr$rkr"rlrmrnror,r(r6r4Z#decision_forest_regression_trainingr#rqr r7r&r'rr0r:rrr8r9r>r?r@rArsrtrurvZoutOfBagErrorR2rwZoutOfBagErrorPredictionrxZoob_prediction_)rCryrzr{r|r~r}rrZ_featuresPerNodeZn_samples_bootstrapZ dfr_algorithmZdfr_trainingResultrur*r*r+_daal_fit_regressorbs      !rc Cst|rtdt||dur,t||}tdrJ|jdkrJtdt t d}| |j rdt dpj|j df|jdud f|jd vd |jd f|jd kd|jdft| dfg}|rtdr|j|ddt|tjtjgd}t|}t|}|jdkr*|jddkr*tjdtddt|d|jd}t|||jdkr\t|d}|jd|_| |jdkd|jdfg}||rt||||d|j |_!|St"t#|j$|||dS)Nrr.msezCriterion 'mse' was deprecated in v1.0 and will be removed in version 1.2. Use `criterion='squared_error'` which is equivalent.z*sklearn.ensemble.RandomForestRegressor.fitrrFr)r squared_errorrzT' criterion is not supported. Only 'mse' and 'squared_error' criteria are supported.rrrrTrrrrrrrErrrr)%rrr%rDrrrr;r<r=rrrprrrrrr$rrrrrrfrrGrrrrrrrrr rrr*r*r+_fit_regressorsl             rcCs^|jd|jkr.td|jdd|jdt|}tj|d}|||j}|j}| S)Nrrz2 features, but RandomForestRegressor is expecting r)rK) rfrgr%rrmZ%decision_forest_regression_predictionrtrvrr)rCryr~Zdfr_algZdfr_predictionResultrr*r*r+_daal_predict_regressors rcCst|}|dur&|tjtjfvr&tj}|dur>tj||d}nvt|tjr\tj|||d}nX|durptjtjg}t |dd|dd}|j dkrt d|j |fkrt d |j |f|S)NrFC) accept_sparserFrGorderrz)Sample weights must be 1D array or scalarz'sample_weight.shape == {}, expected {}!)rr$rrZonesr r&NumberZfullrrr%rfr/)r{ryrGr3r*r*r+rs(    rcseZdZejZedr&dfd d Zndfdd ZedddZefddZ efddZ edre de ddZ e ddZZS)r r.drNrrrrTFrcsRtt|j||||||||| | | | | |||d||_||_||_||_d|_dS)N)rlrrqr8r7r9r(r?r>r^rpn_jobsriverboser class_weight)rr __init__rr4r@rAr:)rCrlrrqr8r7r9r(r?r>r^rprrirrrrr4r@rA __class__r*r+r&s. zRandomForestClassifier.__init__csFtt|j||||||||| | | | | ||||||d||_||_dS)N)rlrrqr8r7r9r(r?r>r:r^rprrirrrrr4)rr rr@rA)rCrlrrqr8r7r9r(r?r>r:r^rprrirrrrr4r@rArr*r+rSs. cCst||||dSa? Build a forest of trees from the training set (X, y). Parameters ---------- X : {array-like, sparse matrix} of shape (n_samples, n_features) The training input samples. Internally, its dtype will be converted to ``dtype=np.float32``. If a sparse matrix is provided, it will be converted into a sparse ``csc_matrix``. y : array-like of shape (n_samples,) or (n_samples, n_outputs) The target values (class labels in classification, real numbers in regression). sample_weight : array-like of shape (n_samples,), default=None Sample weights. If None, then samples are equally weighted. Splits that would create child nodes with net zero or negative weight are ignored while searching for a split in each node. In the case of classification, splits are also ignored if they would result in any single class carrying a negative weight in either child node. Returns ------- self : object r)rrCryrzr{r*r*r+rszRandomForestClassifier.fitcstd}|t|ddft| dfg}t|drV||jdkd|jdfg}||srtt| |St d r|j |d d t |gd t jt jgd }t||S) Predict class for X. The predicted class of an input sample is a vote by the trees in the forest, weighted by their probability estimates. That is, the predicted class is the one with highest mean probability estimate across the trees. Parameters ---------- X : {array-like, sparse matrix} of shape (n_samples, n_features) The input samples. Internally, its dtype will be converted to ``dtype=np.float32``. If a sparse matrix is provided, it will be converted into a sparse ``csr_matrix``. Returns ------- y : ndarray of shape (n_samples,) or (n_samples, n_outputs) The predicted classes. z/sklearn.ensemble.RandomForestClassifier.predictrvoneDAL model was not trained.rrrrrr.FrZcsrZcscZcoorrG)rrhasattrrrrrrr predictrrrr$rrrrCryrrrr*r+rs0    zRandomForestClassifier.predictcs$tdr|j|ddt|drjz t|}WntyFt|}Yn0||jkrjtd|d|jdtd}| t|d d ft | d ft d d fg}t|dr| |j dkd|j dfg}||stt||St|tjtjgd}t|tdr|j|ddt||S)a Predict class probabilities for X. The predicted class probabilities of an input sample are computed as the mean predicted class probabilities of the trees in the forest. The class probability of a single tree is the fraction of samples of the same class in a leaf. Parameters ---------- X : {array-like, sparse matrix} of shape (n_samples, n_features) The input samples. Internally, its dtype will be converted to ``dtype=np.float32``. If a sparse matrix is provided, it will be converted into a sparse ``csr_matrix``. Returns ------- p : ndarray of shape (n_samples, n_classes), or a list of n_outputs such arrays if n_outputs > 1. The class probabilities of the input samples. The order of the classes corresponds to that in the attribute :term:`classes_`. r.Frrgrrrz5sklearn.ensemble.RandomForestClassifier.predict_probarvrr)rriz$oneDAL version is lower than 2021.4.rrrrrz0.23)rrrrr1rrgr%rrrrrrrrr predict_probarr$rrrZ_check_n_featuresr)rCryZ num_featuresrrrr*r+rsJ        z$RandomForestClassifier.predict_probaoAttribute `n_features_` was deprecated in version 1.0 and will be removed in 1.2. Use `n_features_in_` instead.cCs|jSNrgrBr*r*r+rhsz"RandomForestClassifier.n_features_c Csdt|dr|jr|jStdr(t|n t|d|jd}|jd}|j|j|j|j |j |j |j |j dd }tds|j|d<tfi|}g}t|j}t|jD]}t|}|j|ttjjdtdr|j|_n|j|_|j|_||_||_t|j ||} | j| j!| j"| j#d } t$|jtj%|gtj&d |j|_'|j'(| |)|q||_|S) Nrs0.22rvr rrqr8r7r9r(r?r>rir.r:rirq node_countZnodesvaluesr)*rrsrrrrerrqr8r7r9r(r?r>r:r rrirangerlr set_paramsrjr$rkint32r"rgrhrrm getTreeStatervrnode_arvalue_arr arrayintptree_ __setstate__append) rCrreparamsestrrandom_state_checkedrHest_itree_i_state_classtree_i_state_dictr*r*r+rsl            z#RandomForestClassifier._estimators_)rrNrrrrNrTFNNrFNrNrr)rrNrrrrNrNTFNNrFNrNrr)N)__name__ __module__ __qualname__RandomForestClassifier_original__doc__rrrrrrrpropertyrhr __classcell__r*r*rr+r "sv.. ,9 r cseZdZejZedrNdddddddddd d ddd d ddd dd fdd Zn8ddddddddddd d ddd d ddd ddfdd ZedddZefddZ edre de ddZ e ddZ ZS)r r.rrNrrrrTFrr)rrqr8r7r9r(r?r>r^rprrirrrr4r@rAcsPtt|j||||||||| | | | | ||d||_||_||_||_d|_dS)N)rlrrqr8r7r9r(r?r>r^rprrirr)rr rrr4r@rAr:)rCrlrrqr8r7r9r(r?r>r^rprrirrrr4r@rArr*r+r[s, zRandomForestRegressor.__init__r)rrqr8r7r9r(r?r>r:r^rprrirrrr4r@rAcsDtt|j||||||||| | | | | |||||d||_||_dS)N)rlrrqr8r7r9r(r?r>r:r^rprrirrrr4)rr rr@rA)rCrlrrqr8r7r9r(r?r>r:r^rprrirrrr4r@rArr*r+rs, cCst||||dSr)rrr*r*r+rszRandomForestRegressor.fitcstd}|t|ddft| dfg}t|drV||jdkd|jdfg}||srtt| |St d r|j |d d t |gd t jt jgd }t||S)rz.sklearn.ensemble.RandomForestRegressor.predictrvrrrrrrr.Frrr)rrrrrrrrr rrrrr$rrrrrr*r+rs0    zRandomForestRegressor.predictrcCs|jSrrrBr*r*r+rhsz!RandomForestRegressor.n_features_c Cs<t|dr|jr|jStdr(t|n t|d|j|j|j|j|j|j |j |j dd }tdsl|j |d<t fi|}g}t|j}t|jD]}t|}|j|ttjjdtdr|j|_n|j|_|j|_t|j|}|j|j|j |j!d}t"|jtj#d gtj$d |j|_%|j%&||'|q|S) Nrsrrvrr.r:rrrr)(rrsrrrrqr8r7r9r(r?r>r:r rrirrlrrrjr$rkrr"rgrhrrmrrvrrrr rrrrr) rCrrrrrHrrrr*r*r+rsZ         z"RandomForestRegressor._estimators_)r)r)N)rrrRandomForestRegressor_originalrrrrrrrrrhrrr*r*rr+r Wsr,, , r )F)N)N)N)N)N)sH            = c = U B 7