DjQXddlZddlZddlZddlmZddlmcm Z ddl m Z GddejZGddeZGdd eZGd d eZGd d ej$ZGdde j(Ze j,eedZGddeZeZeZeZeZy)N)model)ConvergenceWarningc(eZdZdZfdZdZxZS)_DimReductionRegressionzB A base class for dimension reduction regression methods. c (t|||fi|yN)super__init__selfendogexogkwargs __class__s aC:\Crop_Prediction\Backend\crop-ai-system\venv\Lib\site-packages\statsmodels/regression/dimred.pyr z _DimReductionRegression.__init__s //ctj|j}|j|ddf}||j dz}tj |j ||jdz }tjj|}tjj||j j }||_ ||_ tj|||_yNr)npargsortr rmeandotTshapelinalgcholeskysolvewexog_covxr array_split _split_wexog)r n_sliceiixcovxcovxrs r_prepz_DimReductionRegression._prepsZZ # IIb!e  QVVAYvvacc1~ * ""4( IIOOE133 ' ) )  NN1g6r)__name__ __module__ __qualname____doc__r r' __classcell__rs@rrr s07rrc0eZdZdZddZdZdZ ddZy) SlicedInverseRega0 Sliced Inverse Regression (SIR) Parameters ---------- endog : array_like (1d) The dependent variable exog : array_like (2d) The covariates References ---------- KC Li (1991). Sliced inverse regression for dimension reduction. JASA 86, 316-342. c Ft|dkDrd}tj||jjd|z}|j ||j Dcgc]}|jd}}|j Dcgc]}|jd}}tj|}tj|}tj|j|dddf|z|jz }tjj|\} } tj| } | | } | dd| f} tjj!|j"j| } t%|| | } t'| Scc}wcc}w)z Estimate the EDR space using Sliced Inverse Regression. Parameters ---------- slice_n : int, optional Target number of observations per slice rz1SIR.fit does not take any extra keyword argumentsNeigs)lenwarningswarnrrr'r!rrasarrayrrsumreighrrrDimReductionResultsDimReductionResultsWrapper)r slice_nrmsgr"zmnnmncabjjparamsresultss rfitzSlicedInverseReg.fit6sF v;?EC MM# ))//!$/ 7!%!2!2 3AaffQi 3 3!%!2!2 3AQWWQZ 3 3 ZZ^ JJqMffRTT1QW:?+aeeg5yy~~c"1 ZZ^ bE aeH2%dF;)'224 3s !F Fc|j}|j}|j}|j}d}t j |||j f}t|j D]D}t j|j|dd|f}|t j||zz }Ft j||} tjj| \} } t j| t j| j|j} |j| z } |t j|| | zjdz }|Sr)k_vars_covx _slice_means _slice_propsrreshapendimrangerpen_matr7rqrr)r Apr%r>phvkucovxaq_qdqus r_regularized_objectivez'SlicedInverseReg._regularized_objective[s KKzz        JJq1dii. )tyy! At||Qq!tW-A A A  tQyy||E"1 VVArvvacc244( ) TTBY RVVBb a( ))rc |j}|j}|j}|j}|j}|j }|j ||f}dtj|jjtj|j|z}|j ||f}tj||} tj|| } tj| j| } tjj| } tj||f} tjj| | j}dg||zz}t|D]<}t|D]*}| dz} d| ||f<tj| j| }||jz }tj| tj||  }tjtj|| |}|tj| tj|| jz }|tj| tjj| tj| j|z }||||z|z<-?tjj| tj| j|j}|tj| |jz }t|D]}||ddf}||ddf}t|D]]}t|D]M}tj|tj|||z|z|}|||fxxd||z|zzcc<O_|j!S)Nr)rHrMrIr"rJrKrLrrrOrrinvzerosrrNravel)r rQrRrMr%r"r>rSgrrWcovx2aQQijmqcvftrXrumatfmatchcuirVrTfs r_regularized_gradz"SlicedInverseReg._regularized_gradss KKyyzz,,       IIq$i  t||Q(?@ @ IIq$i tQe$ FF577E " YY]]1  XXq$i iiooa)Vq4x q &A4[ &a1a4vvfhh+r266$#344vvbffT2.4ubffT577&;<<ubiiooad9K&LMM!%1T6A: & &YY__Quww 5 6 "&&#%% %w .A1a4A1a4A1X .t.Aq"&&AdFQJ";r?rDrYcnvrgggnrEs rfit_regularizedz SlicedInverseReg.fit_regularizedsWT v;?GC MM#  ?=> >zz.$7 **Y+))//!$/ZZ # IIb!e  QVVAYvvacc{^^Aw/ !+ ,AaffQi , ,!+ ,AQWWQZ , , ZZ^ JJqMK jjm      XXt{{D12F%'VVD\F1T61T6> "F!!!$,I o%!F%fd.I.I&*&<&)'22A- ,s 2J*J/)rt)r_NrtdgMbP?)r(r)r*r+rFr\rqrrrr/r/%s( #3J0-^IL!c3rr/ceZdZdZdZy)PrincipalHessianDirectionsa Principal Hessian Directions (PHD) Parameters ---------- endog : array_like (1d) The dependent variable exog : array_like (2d) The covariates Returns ------- A model instance. Call `fit` to obtain a results instance, from which the estimated parameters can be obtained. References ---------- KC Li (1992). On Principal Hessian Directions for Data Visualization and Dimension Reduction: Another application of Stein's lemma. JASA 87:420. c |jdd}|j|jjz }|j|jjdz }|r)ddlm}|||j }|j}tjd|||}|t|z}tj|j}tjj||} tjj| \} } tj tj"|  } | | } | dd| f} t%|| | }t'|S)a Estimate the EDR space using PHD. Parameters ---------- resid : bool, optional If True, use least squares regression to remove the linear relationship between each covariate and the response, before conducting PHD. Returns ------- A results instance which can be used to access the estimated parameters. residFr)OLSz i,ij,ik->jkNr1)rvr rr#statsmodels.regression.linear_modelrrFrreinsumr3rwrrreigrabsr9r:)r rryr$rrjcmcxcbrArBrCrDrEs rrFzPrincipalHessianDirections.fits " 7E* JJ* * II q) )  ?Aq AA YY}aA . c!f  VVACC[ YY__R $yy}}R 1 ZZ # bE1b5%dF;)'22rN)r(r)r*r+rFrrrrrs ,'3rrc(eZdZdZfdZdZxZS)SlicedAverageVarianceEstimationa] Sliced Average Variance Estimation (SAVE) Parameters ---------- endog : array_like (1d) The dependent variable exog : array_like (2d) The covariates bc : bool, optional If True, use the bias-corrected CSAVE method of Li and Zhu. References ---------- RD Cook. SAVE: A method for dimension reduction and graphics in regression. http://www.stat.umn.edu/RegGraph/RecentDev/save.pdf Y Li, L-X Zhu (2007). Asymptotics for sliced average variance estimation. The Annals of Statistics. https://arxiv.org/pdf/0708.0462.pdf c ftt| ||fi|d|_d|vr|ddurd|_yyy)NFbcT)r SAVEr rr s rr z(SlicedAverageVarianceEstimation.__init__as@ dD"5$9&9 6>fTld2DG3>rc |jdd}|jjd|z}|j||jDcgc]!}t j |j#}}|jDcgc]}|jd}}|jjd}|jsZd}t||D]9\} } t j|| z } || t j| | zz };|t|z}n@d} |D]} | t j| | z } | t|z} d}|jD]k}||jdz }t|jdD]:}||ddf}t j ||}|t j||z }<m||jjdz}t j|} | | dz z| dz dzdzz }| dz | dz dzdzz }|| z||zz }t j|dt#|zt|z z |z}t j$j'|\}}t j(| }||}|dd|f}t j$j+|j,j|}t/|||}t1|Scc}wcc}w)z Estimate the EDR space. Parameters ---------- slice_n : int Number of observations per slice r;2rr_Nr^r1)rvrrr'r!rrwrrrziprxrr3rrNouterr7rr8rrrr9r:)r rr;r"r=cvnsrRvmwcvxicvavcvnr$rjrorVmk1k2av2rArBrCrDrEs rrFz#SlicedAverageVarianceEstimation.fiths**Y+))//!$/ 7#'#4#4 5abffQSSk 5 5"&"3"3 4Qaggaj 4 4 JJ  Q wwBb"+ +3ffQi#oa"&&c*** + #b'MB B #bffQl" # #b'MBB&& 'q Mqwwqz*'A!Q$AAA"&&A,&B' ' $))//!$ $B Aa!eQ Q/Ba%QUQJN+Br'BG#CQR[3r722S8Byy~~b!1 ZZ^ bE aeH2%dF;)'22[6 4s &K.K3)r(r)r*r+r rFr,r-s@rrrIs.?3rrc"eZdZdZfdZxZS)r9aV Results class for a dimension reduction regression. Notes ----- The `params` attribute is a matrix whose columns span the effective dimension reduction (EDR) space. Some methods produce a corresponding set of eigenvalues (`eigs`) that indicate how much information is contained in each basis direction. c4t|||||_yr)r r r2)r rrDr2rs rr zDimReductionResults.__init__s V  r)r(r)r*r+r r,r-s@rr9r9s rr9ceZdZddiZeZy)r:rDcolumnsN)r(r)r*_attrs _wrap_attrsrrrr:r:s)FKrr:c|j\}}|j}||}d}t|D]} ||} | tj| ||ztj||z z} tj tj | | z|krd}n| j||f} tjj| d\|j||f} tj| jfd} d}|dkDs| | }||}||kr|}|}|dz}|dkDr'|j||f}|||fS)a Minimize a function on a Grassmann manifold. Parameters ---------- params : array_like Starting value for the optimization. fun : function The function to be minimized. grad : function The gradient of fun. maxiter : int The maximum number of iterations. gtol : float Convergence occurs when the gradient norm falls below this value. Returns ------- params : array_like The minimizing value for the objective function. fval : float The smallest achieved value of the objective function. cnvrg : bool True if the algorithm converged to a limit point. Notes ----- `params` is 2-d, but `fun` and `grad` should take 1-d arrays `params.ravel()` as arguments. Reference --------- A Edelman, TA Arias, ST Smith (1998). The geometry of algorithms with orthogonality constraints. SIAM J Matrix Anal Appl. http://math.mit.edu/~edelman/publications/geometry_of_algorithms.pdf FTrctj|zztj|zzz}tj|j Sr)rcossinrrb)tpapa0srVvts rgeoz_grass_opt..geosJrvva!e}$q266!a%='88B66"b>'') )rg@g|=r^) rrbrNrrrzr7rLrsvdr)rDfungradr{r|rRdf0r~rYrgmparamsmrsteprf1rrrVrs @@@@rryrysZL <z)CovarianceReduction.fit..s4<<?:Jrc(j| Sr)rrs rrz)CovarianceReduction.fit..s4::a=.rz/CovReduce optimization did not converge, |g|=%fr1)rrrrrarxryrrbrzr7r4r5rr9llfr:) r rsr{r|rRrrDrr~rrr<rEs ` rrFzCovarianceReduction.fits( IIOOA  HH  XXq!f%F!vvayF1Q3!8 F!F(0J(@'(,.U r  6<<>*AA'BCbHC MM#1 2%dF> )'22r)Ng-C6?) r(r)r*r+r rrrFr,r-s@rrrs%N.66-3rr)r4numpyrpandasrstatsmodels.baserstatsmodels.base.wrapperbasewrapperwrapstatsmodels.tools.sm_exceptionsrModelrr/rrResultsr9ResultsWrapperr:populate_wrapperryrSIRPHDrCORErrrrs"''>7ekk74`3.`3F>3!8>3B^3&=^3B%--&!4!4 0)+Nbb31b3L &r