³ς ’B_Kc@s;defd„ƒYZd„Zedjo eƒndS(tConfusionMatrixcBsSeZdZed„Zd„Zd„Zd„Zeee ed„Z d„Z RS(s| The confusion matrix between a list of reference values and a corresponding list of test values. Entry [M{r},M{t}] of this matrix is a count of the number of times that the reference value M{r} corresponds to the test value M{t}. E.g.: >>> ref = 'DET NN VB DET JJ NN NN IN DET NN'.split() >>> test = 'DET VB VB DET NN NN NN IN DET NN'.split() >>> cm = ConfusionMatrix(ref, test) >>> print cm['NN', 'NN'] 3 Note that the diagonal entries (M{Ri}=M{Tj}) of this matrix corresponds to correct values; and the off-diagonal entries correspond to incorrect values. c s«t|ƒt|ƒjotdƒ‚n|oJt|ƒ‰t|ƒ‰‡‡fd†}tt||ƒd|ƒ}ntt||ƒƒ}td„t|ƒDƒƒ}g}|D](}|g} |D] }| dqΛ~ q·~‰d} xXt||ƒD]G\} } ˆ|| || cd75siic3s!x|]}ˆ||VqWdS(N((RR(t confusion(s2/p/zhu/06/nlp/nltk/nltk/metrics/confusionmatrix.pys IsN(tlent ValueErrortFreqDisttsortedtsettdictt enumeratetziptmaxt_valuest_indicest _confusiont _max_conft_totaltsumtranget_correct( tselft referencettestt sort_by_countRtvaluestindicest_[1]Rt_[2]tmax_conftwtg((RRRs2/p/zhu/06/nlp/nltk/nltk/metrics/confusionmatrix.pyt__init__s*    < #    cCs5|\}}|i|}|i|}|i||S(s† @return: The number of times that value C{li} was expected and value C{lj} was given. @rtype: C{int} (RR(Rt.1tlitljRtj((s2/p/zhu/06/nlp/nltk/nltk/metrics/confusionmatrix.pyt __getitem__Ks   cCsd|i|ifS(Ns (RR(R((s2/p/zhu/06/nlp/nltk/nltk/metrics/confusionmatrix.pyt__repr__Us cCs |iƒS(N(tpp(R((s2/p/zhu/06/nlp/nltk/nltk/metrics/confusionmatrix.pyt__str__Yscs…ˆi}ˆi}|ot|d‡fd†ƒ}n|o|| }n|o+g}|D]}|t|ƒq_~} n8g} tt|ƒƒD]} | t| dƒq–~ } td„| Dƒƒ} d| d} |od}d}d }n2tˆi ƒ}d| d }d |dd }d }x—t| ƒD]‰}|d | d7}xd| D]\}|| t|ƒjo-|||| t|ƒi|dƒ7}q[|d |d7}q[W|d7}q<W|dd| d|dt|ƒf7}x t | |ƒD]ό\}}ˆi |}|| |7}xΘ|D]ΐ}ˆi |}|||djo||7}nC|o%||d|||ˆi 7}n|||||7}||jo1|i d ƒ}|| d||dd}q1|d 7}q1W|d7}qW|dd| d|dt|ƒf7}|d7}|pC|d7}x6t |ƒD]$\}}|d|d|f7}qUWn|S(s• @return: A multi-line string representation of this confusion matrix. @type truncate: int @param truncate: If specified, then only show the specified number of values. Any sorting (e.g., sort_by_count) will be performed before truncation. @param sort_by_count: If true, then sort by the count of each label in the reference data. I.e., labels that occur more frequently in the reference label will be towards the left edge of the matrix, and labels that occur less frequently will be towards the right edge. @todo: add marginals? Rcstˆiˆi|ƒ S((RRR(R(R(s2/p/zhu/06/nlp/nltk/nltk/metrics/confusionmatrix.pytpsicssx|]}t|ƒVqWdS(N(R (RR((s2/p/zhu/06/nlp/nltk/nltk/metrics/confusionmatrix.pys |st%ss | is%5.1f%%s .tdt t.ts |s | s %s-+-%s+ t-igY@ts| s(row = reference; col = test) s Value key: s%6d: %s (RRR tstrRR RRtrjustRRRtrfindR(Rt show_percentstvalues_in_chartttruncateRRRR Rt value_stringsR!tntvaluelent value_formattentrylent entry_formattzerostrtsRR'R(R)t prevspacetvalue((Rs2/p/zhu/06/nlp/nltk/nltk/metrics/confusionmatrix.pyR,\sl  +7  -*   % "*    cCsn|i}d}tt|ƒd ƒ}d| d}x2tt|ƒƒD]}|||||f7}qHW|S(Ns Value key: is %sd: %s (RR R(RRR7tindexlent key_formatR((s2/p/zhu/06/nlp/nltk/nltk/metrics/confusionmatrix.pyR΄s ( t__name__t __module__t__doc__tFalseR%R*R+R-tTruetNoneR,R(((s2/p/zhu/06/nlp/nltk/nltk/metrics/confusionmatrix.pyR s 0  WcCs[diƒ}diƒ}dG|GHdG|GHdGHt||ƒGHt||ƒidtƒGHdS(Ns DET NN VB DET JJ NN NN IN DET NNs DET VB VB DET NN NN NN IN DET NNs Reference =s Test =sConfusion matrix:R(tsplitRR,RM(RR((s2/p/zhu/06/nlp/nltk/nltk/metrics/confusionmatrix.pytdemoΎs    t__main__N(tobjectRRPRI(((s2/p/zhu/06/nlp/nltk/nltk/metrics/confusionmatrix.pys s΅