ó áp7]c@stddlZddlZddlmZddlmZdefd„ƒYZ defd„ƒYZ dd„Z dS( iÿÿÿÿN(tpinv(t utils_oldtContrastResultscBs;eZdZddddddd„Zd„Zd„ZRS(s¸ Results from looking at a particular contrast of coefficients in a parametric model. The class does nothing, it is a container for the results from T and F contrasts. cCsR|dk r*||_||_||_n$||_||_||_||_dS(N(tNonetFtdf_denomtdf_numtttsdteffect(tselfRRRR RR((s?lib/python2.7/site-packages/statsmodels/sandbox/contrast_old.pyt__init__s       cCs!t|dƒr|jS|jSdS(NR(thasattrRR(R ((s?lib/python2.7/site-packages/statsmodels/sandbox/contrast_old.pyt __array__scCset|dƒr/dt|jƒ|j|jfSdt|jƒt|jƒt|jƒ|jfSdS(NRs*s1(R treprRRRR RR(R ((s?lib/python2.7/site-packages/statsmodels/sandbox/contrast_old.pyt__str__ s N(t__name__t __module__t__doc__RR R R(((s?lib/python2.7/site-packages/statsmodels/sandbox/contrast_old.pyRs  tContrastcBsAeZdZdd„Zd„Zd„Zd„ZeeƒZRS(sh This class is used to construct contrast matrices in regression models. They are specified by a (term, formula) pair. The term, T, is a linear combination of columns of the design matrix D=formula(). The matrix attribute is a contrast matrix C so that colspan(dot(D, C)) = colspan(dot(D, dot(pinv(D), T))) where pinv(D) is the generalized inverse of D. Further, the matrix Tnew = dot(C, D) is full rank. The rank attribute is the rank of dot(D, dot(pinv(D), T)) In a regression model, the contrast tests that E(dot(Tnew, Y)) = 0 for each column of Tnew. tcCs=||_||_|dkr0t|ƒ|_n ||_dS(NR(ttermtformulatstrtname(R RRR((s?lib/python2.7/site-packages/statsmodels/sandbox/contrast_old.pyR As    cCs.dtit|jƒd6t|jƒd6ƒS(Ns RR(RRRR(R ((s?lib/python2.7/site-packages/statsmodels/sandbox/contrast_old.pyRIscOsÚtj|jƒ}|jj|_tjtj|||Žƒƒ}|jdkrj|jddf|_nt j |ƒ|_ |jj ||Ž|_ t|j |j ƒ|_y|jjd|_Wnd|_nXdS(s  Construct a contrast matrix C so that colspan(dot(D, C)) = colspan(dot(D, dot(pinv(D), T))) where pinv(D) is the generalized inverse of D=self.D=self.formula(). If the design, self.D is already set, then evaldesign can be set to False. iiN(tcopyRRt namespacetnpt transposetarraytndimtshapetutilstclean0tTtdesigntDtcontrastfromcolst_matrixtmatrixtrank(R targstkwRR"((s?lib/python2.7/site-packages/statsmodels/sandbox/contrast_old.pytcompute_matrixMs !cCs#t|dƒs|jƒn|jS(s` This will fail if the formula needs arguments to construct the design. R&(R R+R&(R ((s?lib/python2.7/site-packages/statsmodels/sandbox/contrast_old.pyt _get_matrixjs ( RRRR RR+R,tpropertyR'(((s?lib/python2.7/site-packages/statsmodels/sandbox/contrast_old.pyR)s     cCs^tj|ƒ}tj|ƒ}|j\}}|jd|krb|jd|krbtdƒ‚n|dkr}t|ƒ}n|jd|kr¨tj||ƒj}n*|}tj|tj||jƒƒj}tj||jƒ}t|jƒdkr|df|_nt j |ƒ|jdkrQt j |ƒ}tj||ƒj}ntj |ƒS(sV From an n x p design matrix D and a matrix L, tries to determine a p x q contrast matrix C which determines a contrast of full rank, i.e. the n x q matrix dot(transpose(C), pinv(D)) is full rank. L must satisfy either L.shape[0] == n or L.shape[1] == p. If L.shape[0] == n, then L is thought of as representing columns in the column space of D. If L.shape[1] == p, then L is thought of as what is known as a contrast matrix. In this case, this function returns an estimable contrast corresponding to the dot(D, L.T) Note that this always produces a meaningful contrast, not always with the intended properties because q is always non-zero unless L is identically 0. That is, it produces a contrast that spans the column space of L (after projection onto the column space of D). iisshape of L and D mismatchedN( RtasarrayRt ValueErrorRRtdotR"tlenR R(tfullranktsqueeze(tLR$tpseudotntptCtLp((s?lib/python2.7/site-packages/statsmodels/sandbox/contrast_old.pyR%ts$& $( RtnumpyRt numpy.linalgRtstatsmodels.sandboxRR tobjectRRRR%(((s?lib/python2.7/site-packages/statsmodels/sandbox/contrast_old.pyts  "K