a )a;9@sddlZddlZddlmZddlmZddlmZddl m Z ddl m Z ddl Z ddlmZmZmZmZdd lmZddlZed rdd lmZn dd lmZed rdd lmZn"ed rddlmZn ddlmZGdddeZdS)N)sqrt)issparse) check_array)check_is_fitted) stable_cumsum) getFPTypeget_patch_messagesklearn_check_versionPatchingConditionsChain)support_usm_ndarray0.22)PCA0.23)_infer_dimension)_infer_dimension_c@sdeZdZejZdddZdd Zd d Zd d ZddZ dddZ e ddZ e dddZ dS)rNTFautocCs.||_||_||_||_||_||_||_dS)N) n_componentscopywhiten svd_solvertoliterated_power random_state)selfrrrrrrrrAlib/python3.9/site-packages/daal4py/sklearn/decomposition/_pca.py__init__0s z PCA.__init__cCsz|dkr||krvtdn\d|kr4t||ksNntd|t||fn(|dkrvt|tjsvtd|t|fdS)Nmlez?n_components='mle' is only supported if n_samples >= n_featuresrzZn_components=%r must be between 0 and min(n_samples, n_features)=%r with svd_solver='full'zSn_components=%r must be of type int when greater than or equal to 1, was of type=%r) ValueErrormin isinstancenumbersZIntegraltype)rr n_samples n_featuresrrr_validate_n_componentsBs    zPCA._validate_n_componentscs|j\}}t||}|dkr"|}n|dkr0|}n|}t|}tj|dd}||} | j|_| jt fddt |D} tj |ddd |d } | |} | j } t | jd }|}||}|dkrtd rt||}n t|||}n8d |kr d kr,nnt|}t j||ddd}||kr|jd |krZ||d|_n*| }||d|8}||||_nd|_|||_|_| d||_||_|d||_|d||_t |d|j|_dS)Nrr ZcovarianceMatrix)fptypeZoutputMatrixTypecsg|]}||fqSrr).0i covariancerr fz)PCA._fit_full_daal4py..ZcorrelationDense eigenvalueT)r)methodZresultsToComputeZisDeterministicZ nComponentsrr?rightZsider)shaper"rdaal4pyr-computemeanZravelmean_nparrayrangeZpcaZ eigenvectorsZmaximumZ eigenvaluessumr rrr searchsortednoise_variance_ n_samples_ n_features_ components_ n_components_explained_variance_explained_variance_ratio_rsingular_values_)rXrr&r'Zn_sf_minZdaal_n_componentsfpTypeZcovariance_algoZcovariance_resZ variances_Zpca_algZpca_resrBrDZtot_varrE ratio_cumsumZ resid_var_rr,r_fit_full_daal4pySsh        zPCA._fit_full_daal4pyc Cs&|j\}}||||||t|jd}|j}|j}|dkrhtdrXt|j|}qt |j||}n6d|kr|dkrnnt |j }t j ||ddd}|t||kr|j|d|_nd|_|||_|_|jd||_||_|jd||_|j d||_ |jd||_|||fS) Nrrrr2r3r4r r)r5r(rJr"rBrFr rrDrrrEr:r>r8r?r@rArC) rrGrr&r'UVSrIrrr _fit_fulls8   z PCA._fit_fullc Cs$t|rtdtdr4|j|tjtjgddd}nt|tjtjgddd}|jdur||j dkrlt |j }qt |j d}n|j}|j |_ |j d|j dd k}|j d krL|d krd |_ n|j d|j d|}}}t d |||gd|||gd|d gg}|dkrFt|dddf|dddfdkrFd|_ nd |_ |r^|j d krtdrz|j||jd}nt||jd}td}||j d kd|j dfg} | r||dfg} | r|||} nt|||} n.|j dvr||||j } ntd|j || S)NzOPCA does not support sparse input. See TruncatedSVD for a possible alternative.rTF)dtypeZ ensure_2drarpackr rrrrfullgK=g¥gYi"`> randomized)rzsklearn.decomposition.PCA.fit'z?' SVD solver is not supported. Only 'full' solver is supported.z`The shape of X does not satisfy oneDAL requirements: number of features / number of samples >= 2)rPrRzUnrecognized svd_solver='{0}')r TypeErrorr Z_validate_datar:float64float32rrrr"r5Z_fit_svd_solverr;dotrr and_conditionsrN PCA_originalZ_fit_truncatedr!format write_log) rrGrZshape_good_for_daalnpkZregression_coefs_patching_status _dal_readyresultrrr_fitsp      &   zPCA._fitcCs(tdrt|nt|ddgtdtdr:|j|ddt|tjtjg|d}t|}t }|j durx|j d |d <|r|r|j d |j d |d <q|j d |d <n,|rtjd |j jd f|j d|jd|d <|jd |jkr td|jd d|jdtj|d||j|}|jS)Nr r9rB)Z all_or_anyz1.0F)reset)rOZforce_all_finite)r r8r r0rr2rOzX has z features, but PCA is expecting z features as input)r))r rallZ_check_feature_namesrr:rUrVrdictr9Zreshaper@rDrQr5rOrAr!r6Z pca_transformr7rBZtransformedData)rrGrscale_eigenvaluescheck_XrHZtr_dataZtr_resrrr_transform_daal4pys@   zPCA._transform_daal4pycCsJtd}||jdkdfg}||r>|j||jdddSt||S)a Apply dimensionality reduction to X. X is projected on the first principal components previously extracted from a training set. Parameters ---------- X : array-like of shape (n_samples, n_features) New data, where `n_samples` is the number of samples and `n_features` is the number of features. Returns ------- X_new : array-like of shape (n_samples, n_components) Projection of X in the first principal components, where `n_samples` is the number of samples and `n_components` is the number of the components. z#sklearn.decomposition.PCA.transformrNumber of components <= 0.TFrrirh)r rXrCr[rjrrY transform)rrGr_r`rrrrm&s  z PCA.transformc Cs||\}}}td}||dudfg}|rx||jdkdfg}|r`|j||jddd}qtj|jdf|j d}nJ|ddd|jf}|jr|t |j dd 9}n||d|j9}|}| |S) ap Fit the model with X and apply the dimensionality reduction on X. Parameters ---------- X : array-like of shape (n_samples, n_features) Training data, where `n_samples` is the number of samples and `n_features` is the number of features. y : Ignored Ignored. Returns ------- X_new : ndarray of shape (n_samples, n_components) Transformed values. Notes ----- This method returns a Fortran-ordered array. To convert it to a C-ordered array, use 'np.ascontiguousarray'. z'sklearn.decomposition.PCA.fit_transformNzStock fitting was used.rrkFrlrer ) rbr rXrCrjrr:emptyr@rOrr5r[) rrGyrKrM_r_r`rarrr fit_transformFs.   zPCA.fit_transform)NTFrrrN)FTT)N)__name__ __module__ __qualname__rY__doc__rr(rJrNrbrjr rmrqrrrrr-s$ ?#M $ r)Znumpyr:r$ZmathrZ scipy.sparserZ sklearn.utilsrZsklearn.utils.validationrZsklearn.utils.extmathrr6Z_utilsrr r r Z_device_offloadr ZloggingZsklearn.decomposition._pcarrYZsklearn.decomposition.pcarrrrrrs&