ó ‡ˆ\c@ s…dZddlmZddlZddlmZddlZddlm Z ddl m Z ddl m Z mZmZmZdd lmZmZdd lmZmZmZmZdd lmZmZdd lmZdd lmZddl m!Z!ddl"m#Z#ddl$m%Z%de efd„ƒYZ&de'fd„ƒYZ(dd„Z*de efd„ƒYZ+e,dd„Z-dS(s'Calibration of predicted probabilities.iÿÿÿÿ(tdivisionN(tlog(t fmin_bfgs(t LabelEncoderi(t BaseEstimatortClassifierMixintRegressorMixintclone(tlabel_binarizetLabelBinarizer(t check_X_yt check_arrayt indexablet column_or_1d(tcheck_is_fittedtcheck_consistent_length(t signature(tIsotonicRegression(t LinearSVC(tcheck_cv(t'_check_binary_probabilistic_predictionstCalibratedClassifierCVcB s>eZdZdddd„Zdd„Zd„Zd„ZRS(sÜ Probability calibration with isotonic regression or sigmoid. See glossary entry for :term:`cross-validation estimator`. With this class, the base_estimator is fit on the train set of the cross-validation generator and the test set is used for calibration. The probabilities for each of the folds are then averaged for prediction. In case that cv="prefit" is passed to __init__, it is assumed that base_estimator has been fitted already and all data is used for calibration. Note that data for fitting the classifier and for calibrating it must be disjoint. Read more in the :ref:`User Guide `. Parameters ---------- base_estimator : instance BaseEstimator The classifier whose output decision function needs to be calibrated to offer more accurate predict_proba outputs. If cv=prefit, the classifier must have been fit already on data. method : 'sigmoid' or 'isotonic' The method to use for calibration. Can be 'sigmoid' which corresponds to Platt's method or 'isotonic' which is a non-parametric approach. It is not advised to use isotonic calibration with too few calibration samples ``(<<1000)`` since it tends to overfit. Use sigmoids (Platt's calibration) in this case. cv : integer, cross-validation generator, iterable or "prefit", optional Determines the cross-validation splitting strategy. Possible inputs for cv are: - None, to use the default 3-fold cross-validation, - integer, to specify the number of folds. - :term:`CV splitter`, - An iterable yielding (train, test) splits as arrays of indices. For integer/None inputs, if ``y`` is binary or multiclass, :class:`sklearn.model_selection.StratifiedKFold` is used. If ``y`` is neither binary nor multiclass, :class:`sklearn.model_selection.KFold` is used. Refer :ref:`User Guide ` for the various cross-validation strategies that can be used here. If "prefit" is passed, it is assumed that base_estimator has been fitted already and all data is used for calibration. .. versionchanged:: 0.20 ``cv`` default value if None will change from 3-fold to 5-fold in v0.22. Attributes ---------- classes_ : array, shape (n_classes) The class labels. calibrated_classifiers_ : list (len() equal to cv or 1 if cv == "prefit") The list of calibrated classifiers, one for each crossvalidation fold, which has been fitted on all but the validation fold and calibrated on the validation fold. References ---------- .. [1] Obtaining calibrated probability estimates from decision trees and naive Bayesian classifiers, B. Zadrozny & C. Elkan, ICML 2001 .. [2] Transforming Classifier Scores into Accurate Multiclass Probability Estimates, B. Zadrozny & C. Elkan, (KDD 2002) .. [3] Probabilistic Outputs for Support Vector Machines and Comparisons to Regularized Likelihood Methods, J. Platt, (1999) .. [4] Predicting Good Probabilities with Supervised Learning, A. Niculescu-Mizil & R. Caruana, ICML 2005 tsigmoidtwarncC s||_||_||_dS(N(tbase_estimatortmethodtcv(tselfRRR((s2lib/python2.7/site-packages/sklearn/calibration.pyt__init__ls  cC s t||ddddgdtƒ\}}t||ƒ\}}tƒj|ƒ}|j|_t|jtƒrx|jn!t |jdƒr–|jj nd}|rõt j g|jD]!}t j||kƒ|k^q²ƒrõtd||fƒ‚ng|_|jdkrtdd ƒ}n |j}|jd kr‘t|d |jƒ}|dk rn|j|||ƒn|j||ƒ|jj|ƒn‹t|j|d tƒ} t|jƒj} t|ƒj} |dk rüd | krütjd| ƒd} n4|dk r*t|dtƒ}t||ƒn|} xé| j ||ƒD]Õ\} }t!|ƒ}| dk rŒ|j|| || d | | ƒn|j|| || ƒt|d |jd|jƒ}|dk rð|j||||||ƒn|j||||ƒ|jj|ƒqCW|S(sÉFit the calibrated model Parameters ---------- X : array-like, shape (n_samples, n_features) Training data. y : array-like, shape (n_samples,) Target values. sample_weight : array-like, shape = [n_samples] or None Sample weights. If None, then samples are equally weighted. Returns ------- self : object Returns an instance of self. t accept_sparsetcsctcsrtcootforce_all_finitetn_foldss^Requesting %d-fold cross-validation but provided less than %d examples for at least one class.t random_stateitprefitRt classifiert sample_weights\%s does not support sample_weight. Samples weights are only used for the calibration itself.t ensure_2dtclassesN("R tFalseR R tfittclasses_t isinstanceRtintthasattrR"tNonetnptanytsumt ValueErrortcalibrated_classifiers_RRt_CalibratedClassifierRtappendRtTrueRt parametersttypet__name__twarningsRR RtsplitR(RtXtyR&tleR"tclass_Rtcalibrated_classifierRtfit_parameterstestimator_nametbase_estimator_sample_weightttrainttesttthis_estimator((s2lib/python2.7/site-packages/sklearn/calibration.pyR*qsf $ 1            cC st|ddgƒt|ddddgdtƒ}tj|jdt|jƒfƒ}x*|jD]}|j |ƒ}||7}qcW|t|jƒ}|S( s£Posterior probabilities of classification This function returns posterior probabilities of classification according to each class on an array of test vectors X. Parameters ---------- X : array-like, shape (n_samples, n_features) The samples. Returns ------- C : array, shape (n_samples, n_classes) The predicted probas. R+R4RRRR R!i( RR R)R0tzerostshapetlenR+R4t predict_proba(RR=t mean_probaRAtproba((s2lib/python2.7/site-packages/sklearn/calibration.pyRKÉs %cC s6t|ddgƒ|jtj|j|ƒddƒS(sVPredict the target of new samples. Can be different from the prediction of the uncalibrated classifier. Parameters ---------- X : array-like, shape (n_samples, n_features) The samples. Returns ------- C : array, shape (n_samples,) The predicted class. R+R4taxisi(RR+R0targmaxRK(RR=((s2lib/python2.7/site-packages/sklearn/calibration.pytpredictçsN(R:t __module__t__doc__R/RR*RKRP(((s2lib/python2.7/site-packages/sklearn/calibration.pyRs M X R5cB s;eZdZddd„Zd„Zdd„Zd„ZRS(s”Probability calibration with isotonic regression or sigmoid. It assumes that base_estimator has already been fit, and trains the calibration on the input set of the fit function. Note that this class should not be used as an estimator directly. Use CalibratedClassifierCV with cv="prefit" instead. Parameters ---------- base_estimator : instance BaseEstimator The classifier whose output decision function needs to be calibrated to offer more accurate predict_proba outputs. No default value since it has to be an already fitted estimator. method : 'sigmoid' | 'isotonic' The method to use for calibration. Can be 'sigmoid' which corresponds to Platt's method or 'isotonic' which is a non-parametric approach based on isotonic regression. classes : array-like, shape (n_classes,), optional Contains unique classes used to fit the base estimator. if None, then classes is extracted from the given target values in fit(). See also -------- CalibratedClassifierCV References ---------- .. [1] Obtaining calibrated probability estimates from decision trees and naive Bayesian classifiers, B. Zadrozny & C. Elkan, ICML 2001 .. [2] Transforming Classifier Scores into Accurate Multiclass Probability Estimates, B. Zadrozny & C. Elkan, (KDD 2002) .. [3] Probabilistic Outputs for Support Vector Machines and Comparisons to Regularized Likelihood Methods, J. Platt, (1999) .. [4] Predicting Good Probabilities with Supervised Learning, A. Niculescu-Mizil & R. Caruana, ICML 2005 RcC s||_||_||_dS(N(RRR((RRRR(((s2lib/python2.7/site-packages/sklearn/calibration.pyR$s  cC sát|jƒ}t|jdƒra|jj|ƒ}|jdkr¿|dd…tjf}q¿n^t|jdƒr³|jj|ƒ}|dkr¿|dd…dd…f}q¿n t dƒ‚|j j |jjƒ}||fS(Ntdecision_functioniRKis<classifier has no decision_function or predict_proba method.( RJR+R.RRStndimR0tnewaxisRKt RuntimeErrortlabel_encoder_t transform(RR=t n_classestdft idx_pos_class((s2lib/python2.7/site-packages/sklearn/calibration.pyt_preproc)s "  c C s)tƒ|_|jdkr.|jj|ƒn|jj|jƒ|jj|_t||jƒ}|j|ƒ\}}g|_x¢t ||j ƒD]Ž\}}|j dkrÀt ddƒ} n.|j dkrÛt ƒ} ntd|j ƒ‚| j||dd…|f|ƒ|jj| ƒq“W|S(sËCalibrate the fitted model Parameters ---------- X : array-like, shape (n_samples, n_features) Training data. y : array-like, shape (n_samples,) Target values. sample_weight : array-like, shape = [n_samples] or None Sample weights. If None, then samples are equally weighted. Returns ------- self : object Returns an instance of self. tisotonict out_of_boundstclipRs1method should be "sigmoid" or "isotonic". Got %s.N(RRWR(R/R*R+RR\t calibrators_tziptTRRt_SigmoidCalibrationR3R6( RR=R>R&tYRZR[tktthis_dft calibrator((s2lib/python2.7/site-packages/sklearn/calibration.pyR*<s$    #c C s>t|jƒ}tj|jd|fƒ}|j|ƒ\}}xat||j|jƒD]G\}}}|dkr|d7}n|j |ƒ|dd…|fgrÄZ| ð?( RJR+R0RHRIR\RaRbR`RPR2RUtisnan( RR=RYRMRZR[ReRfRg((s2lib/python2.7/site-packages/sklearn/calibration.pyRKis%  # -,N(R:RQRRR/RR\R*RK(((s2lib/python2.7/site-packages/sklearn/calibration.pyR5ùs *  -c  s6t|ƒ}t|ƒ}|‰tjtjƒj‰ttj|dkƒƒ}|jd|}tj|jƒ‰|d|dˆ|dkR&tprior0tprior1RqRwtAB0tAB_((RnRbRoR&Rps2lib/python2.7/site-packages/sklearn/calibration.pyt_sigmoid_calibration“s     'RccB s#eZdZdd„Zd„ZRS(s†Sigmoid regression model. Attributes ---------- a_ : float The slope. b_ : float The intercept. cC sOt|ƒ}t|ƒ}t||ƒ\}}t|||ƒ\|_|_|S(sÒFit the model using X, y as training data. Parameters ---------- X : array-like, shape (n_samples,) Training data. y : array-like, shape (n_samples,) Training target. sample_weight : array-like, shape = [n_samples] or None Sample weights. If None, then samples are equally weighted. Returns ------- self : object Returns an instance of self. (R R R€ta_tb_(RR=R>R&((s2lib/python2.7/site-packages/sklearn/calibration.pyR*às   cC s/t|ƒ}ddtj|j||jƒS(sPredict new data by linear interpolation. Parameters ---------- T : array-like, shape (n_samples,) Data to predict from. Returns ------- T_ : array, shape (n_samples,) The predicted data. gð?(R R0RiRR‚(RRb((s2lib/python2.7/site-packages/sklearn/calibration.pyRPús N(R:RQRRR/R*RP(((s2lib/python2.7/site-packages/sklearn/calibration.pyRcÕs  ic C sMt|ƒ}t|ƒ}|rE||jƒ|jƒ|jƒ}n3|jƒdksi|jƒdkrxtdƒ‚nt||ƒ}tjdd |dƒ}tj||ƒd}tj|d|dt |ƒƒ}tj|d|dt |ƒƒ}tj|dt |ƒƒ}|dk} || || } || || } | | fS( s¶Compute true and predicted probabilities for a calibration curve. The method assumes the inputs come from a binary classifier. Calibration curves may also be referred to as reliability diagrams. Read more in the :ref:`User Guide `. Parameters ---------- y_true : array, shape (n_samples,) True targets. y_prob : array, shape (n_samples,) Probabilities of the positive class. normalize : bool, optional, default=False Whether y_prob needs to be normalized into the bin [0, 1], i.e. is not a proper probability. If True, the smallest value in y_prob is mapped onto 0 and the largest one onto 1. n_bins : int Number of bins. A bigger number requires more data. Bins with no data points (i.e. without corresponding values in y_prob) will not be returned, thus there may be fewer than n_bins in the return value. Returns ------- prob_true : array, shape (n_bins,) or smaller The true probability in each bin (fraction of positives). prob_pred : array, shape (n_bins,) or smaller The mean predicted probability in each bin. References ---------- Alexandru Niculescu-Mizil and Rich Caruana (2005) Predicting Good Probabilities With Supervised Learning, in Proceedings of the 22nd International Conference on Machine Learning (ICML). See section 4 (Qualitative Analysis of Predictions). iis?y_prob has values outside [0, 1] and normalize is set to False.ggð?g:Œ0âŽyE>tweightst minlengthgÜ1¯ð?( R tmintmaxR3RR0tlinspacetdigitizetbincountRJ( ty_truety_probt normalizetn_binstbinstbinidstbin_sumstbin_truet bin_totaltnonzerot prob_truet prob_pred((s2lib/python2.7/site-packages/sklearn/calibration.pytcalibration_curve s *  '$!! (.RRt __future__RR;tmathRtnumpyR0tscipy.optimizeRtsklearn.preprocessingRtbaseRRRRt preprocessingRR tutilsR R R R tutils.validationRRt utils.fixesRR]RtsvmRtmodel_selectionRtmetrics.classificationRRtobjectR5R/R€RcR)R–(((s2lib/python2.7/site-packages/sklearn/calibration.pyts(   ""Ûš B6