Dj\ ddlZGddZy)NcteZdZdZdZdZddZdZdZdZ d Z d Z dd Z dd Z d ZdZddZddZddZy)Pcaz A basic class for Principal Component Analysis (PCA). p is the number of dimensions, while N is the number of data points )rgbcymkc|j}|tj|dz }|tj|dz }||_||_d|_y)Nraxis)AnpmeanstdMN_eig)selfrrrs [C:\Crop_Prediction\Backend\crop-ai-system\venv\Lib\site-packages\statsmodels/sandbox/pca.py__calcz Pca.__calc sJ FF BGGA1   BFF1!   Nc Btj|j}|j\}}||c|_|_||kDrddlm}|dt||_ |j|_ |jtj|jt|dz t!|jz dzd||_|t!||k7r t#d|d|_yt%|Dcgc] }t'|c}|_ycc}w)z$ p X N matrix input r)warnzp > n - intentional?Nznames must match data dimension)rarrayTshapenpwarningsrRuntimeWarningrcopy_origA _Pca__calctile_colorsintlen ValueErrortuplestrnames)rdatar.rr r!rxs r__init__z Pca.__init__s HHTN  gg!! tv q5 % ' 8FFH  ggdll3!S5F/F+G+IJ2AN  Uq>? ?"]T u6M!s1v6M0N 6Ms=DcTtj|jjS)z? returns the covariance matrix for the dataset )rcovrrrs rgetCovarianceMatrixzPca.getCovarianceMatrix+svvdffhhrc|jftjj|j }tj |dddd}|d||ddd|ff}||_|jS)zQ returns a tuple of (eigenvalues,eigenvectors) for the data set. Nrr)rrlinalgeigr5argsort)rressortis rgetEigensystemzPca.getEigensystem1ss 99 ))-- 8 8 :;C**SV$TrT*EQs1vag/CDIyyrc(|jdS)Nrr=r4s rgetEigenvalueszPca.getEigenvalues<""$Q''rc(|jdS)Nrr?r4s rgetEigenvectorszPca.getEigenvectors?rArcR|j}|tj|z S)z= "energies" are just normalized eigenvectors )r@rsum)rvs r getEnergieszPca.getEnergiesBs$    {rc ddlm}|jdd|f|jdd|f}}|r|j|j |||j \}}|j \} } |j\} } | | z | | z }} t||j|jD]6\}}}|jdd|||z|||zd| |zdz dzz||8|jI|jd|j|zdz|jd|j|zdzyy) z Generates a 2-dimensional plot of the data set and principle components using matplotlib. ix specifies which p-dimension to put on the x-axis of the plot and iy specifies which to put on the y-axis (0-indexed) rNg?g?) head_widthfcec$z/\sigma$)matplotlib.pyplotpyplotrclfscatterr=xlimylimziprr(arrowr.xlabelylabel)rixiyrPpltr0r valsevsxlxuylyudxdyvalvecrs rplot2dz Pca.plot2dIs5 ( FF1R4L"!  GGI Aa$$&Shhj2hhj2"ur"u2T#%% 5 \ICA IIa#c"g+c#b'kT2b57S.=PTUYZI [ \ :: ! JJs4::b>)+5 6 JJs4::b>)+5 6 "rc :ddlmcm}|r|jt j d}|j |jz}|j|||||||||d|j|jdd|f|jdd|f|jdd|fd|jrE|j|j|dz|j|dz|j|dzy|jy) z Generates a 3-dimensional plot of the data set and principle components using mayavi. ix, iy, and iz specify which of the input p-dimensions to place on each of the x,y,z axes, respectively (0-indexed). rN) scale_factorg333333?z/sigma)rVrWzlabel) enthought.mayavi.mlabmayavimlabrPrzerosrCr@quiver3dpoints3drr.axes)rrXrYizrPrz3rFs rplot3dz Pca.plot3dcs *)  EEG 88A;    !$"5"5"7 7 2bAbE!B%"1 = 466!B$<qt TVVAbD\s K :: FF$**R.1B9PX\XbXbceXfgoXoF p FFHrctj|r/|tj|jjdz}|tj |jdz}|jjd}tj tj|j|kd}|j||_|j|t|z S)a  clips out all data points that are more than a certain number of standard deviations from the mean. sigs can be either a single value or a length-p sequence that specifies the number of standard deviations along each of the p dimensions. rr r) risscalaronesrrrallabsrr&rE)rsigsr r s rsigclipz Pca.sigclipws ;;t bggdffll1o..DBFF466q)) FFLLO FF266$&&>D(a 0vvay Qxrcb|jj|_|jyN)r%r$rr&r4s rresetz Pca.resets!!# rct|||fDcgc]}|duc}}|dk(r td}ni|dkDr td||j|kD}nC| td|}n4|'t j |j|k}n t d||jj}nUt j|d}|jjjd|jdk7r tdt j|jj|z}||jScc}w) aD projects the normalized values onto the components enthresh, nPCs, and cumen determine how many PCs to use if vals is None, the normalized data vectors are the values to project. Otherwise, it should be convertable to a p x N array returns n,p(>threshold) dimension array Nrrz&cannot specify more than one thresholdzShould be unreachableF)r$zshape for vals does not match) rEslicer+energiesrcumsum RuntimeErrorrrrrmatrixrC) rr[enthreshnPCscumenenonnonesr projs rprojectz Pca.projects$/FG! GH q=d A \EF F#MMOh.!$t$"IIdmmo.%7"#:;; <6688D88De,Dvvxx~~a DJJqM1 !@AAyy--/02247Awyy-Hs Ectj|}|j\}}|jjd}||kDr t dtj j tj|jj}tj||f}||ddd|f<||jz}|r)tj|jjStj|jd} tj|jd} tj|j| z| zjS)zP input is an n X q array, where q <= p output is p X n rzq > pNrr )r atleast_2drrr+r8invrrCrrnrrrr) rrnormedr qr!evinvzsrmnssdss r deprojectz Pca.deprojects -- gg! FFLLO q5W% %iimmBIId&:&:&<=??@ XXqe_1RaR4RTTz 88DFF#%% %A&Ctvv1%CHHTVV$S(,// /rc| |j}n@|j}|jd|jjdk7r td|j }t j |}|dd|f|dd|f<|j|d}|j|z }|jt j|jdz}|t j|jdzS)z pc can be a scalar or any sequence of pc indecies if vals is None, the source data is self.A, else whatever is in vals (which must be p x m) Nrz1vals do not have the correct number of componentsFrr ) rrrr+rr zeros_likerrrr)rpcr[pcszpcsupcrBs r subtractPCzPca.subtractPCs <66D66Dzz!}tvv||A. !TUU LLN ]]3 qt9QrT NN4 & FF3J CCtvv1% %Q'''rr})rrT)rrT)NNNN)T)__name__ __module__ __qualname____doc__r(r&r1r5r=r@rCrGrertr{r~rrrrrrrsV *GO* ((74($ !F04(rr)numpyrrrrrrs[([(r