U !`7@sddlZddlmZddlmZddlZddlZddlm Z m Z m Z ddl m Z mZddlmZmZddlmZmZd d d gZd d Zedddd ddd ZGdd d e e e ZdS)N) interpolate) spearmanr) BaseEstimatorTransformerMixinRegressorMixin) check_arraycheck_consistent_length)_check_sample_weight_deprecate_positional_args)'_inplace_contiguous_isotonic_regression _make_uniquecheck_increasingisotonic_regressionIsotonicRegressionc Cst||\}}|dk}|dkrt|dkrdtd|d|}dtt|d}t|d|}t|d|}t|t|krt d|S) aGDetermine whether y is monotonically correlated with x. y is found increasing or decreasing with respect to x based on a Spearman correlation test. Parameters ---------- x : array-like of shape (n_samples,) Training data. y : array-like of shape (n_samples,) Training target. Returns ------- increasing_bool : boolean Whether the relationship is increasing or decreasing. Notes ----- The Spearman correlation coefficient is estimated from the data, and the sign of the resulting estimate is used as the result. In the event that the 95% confidence interval based on Fisher transform spans zero, a warning is raised. References ---------- Fisher transformation. Wikipedia. https://en.wikipedia.org/wiki/Fisher_transformation r)g?g?rrg\(\?zwConfidence interval of the Spearman correlation coefficient spans zero. Determination of ``increasing`` may be suspect.) rlenmathlogZsqrtZtanhnpZsignwarningswarn) xyZrho_Zincreasing_boolFZF_seZrho_0Zrho_1r/lib/python3.8/site-packages/sklearn/isotonic.pyrs" T sample_weighty_miny_max increasingcCs|rtjddntjddd}t|dtjtjgd}tj|||jd}t|||jd}t||}t |||dk s|dk r|dkrtj }|dkrtj }t ||||||S)aSolve the isotonic regression model. Read more in the :ref:`User Guide `. Parameters ---------- y : array-like of shape (n_samples,) The data. sample_weight : array-like of shape (n_samples,), default=None Weights on each point of the regression. If None, weight is set to 1 (equal weights). y_min : float, default=None Lower bound on the lowest predicted value (the minimum value may still be higher). If not set, defaults to -inf. y_max : float, default=None Upper bound on the highest predicted value (the maximum may still be lower). If not set, defaults to +inf. increasing : bool, default=True Whether to compute ``y_`` is increasing (if set to True) or decreasing (if set to False) Returns ------- y_ : list of floats Isotonic fit of y. References ---------- "Active set algorithms for isotonic regression; A unifying framework" by Michael J. Best and Nilotpal Chakravarti, section 3. NF) ensure_2ddtyper&) rZs_rfloat64float32arrayr&r Zascontiguousarrayr infclip)rr r!r"r#orderrrrrOs&" cseZdZdZedddddddZdd Zd d Zdd d ZdddZ ddZ ddZ fddZ fddZ ddZZS)rau Isotonic regression model. Read more in the :ref:`User Guide `. .. versionadded:: 0.13 Parameters ---------- y_min : float, default=None Lower bound on the lowest predicted value (the minimum value may still be higher). If not set, defaults to -inf. y_max : float, default=None Upper bound on the highest predicted value (the maximum may still be lower). If not set, defaults to +inf. increasing : bool or 'auto', default=True Determines whether the predictions should be constrained to increase or decrease with `X`. 'auto' will decide based on the Spearman correlation estimate's sign. out_of_bounds : {'nan', 'clip', 'raise'}, default='nan' Handles how `X` values outside of the training domain are handled during prediction. - 'nan', predictions will be NaN. - 'clip', predictions will be set to the value corresponding to the nearest train interval endpoint. - 'raise', a `ValueError` is raised. Attributes ---------- X_min_ : float Minimum value of input array `X_` for left bound. X_max_ : float Maximum value of input array `X_` for right bound. X_thresholds_ : ndarray of shape (n_thresholds,) Unique ascending `X` values used to interpolate the y = f(X) monotonic function. .. versionadded:: 0.24 y_thresholds_ : ndarray of shape (n_thresholds,) De-duplicated `y` values suitable to interpolate the y = f(X) monotonic function. .. versionadded:: 0.24 f_ : function The stepwise interpolating function that covers the input domain ``X``. increasing_ : bool Inferred value for ``increasing``. Notes ----- Ties are broken using the secondary method from de Leeuw, 1977. References ---------- Isotonic Median Regression: A Linear Programming Approach Nilotpal Chakravarti Mathematics of Operations Research Vol. 14, No. 2 (May, 1989), pp. 303-308 Isotone Optimization in R : Pool-Adjacent-Violators Algorithm (PAVA) and Active Set Methods de Leeuw, Hornik, Mair Journal of Statistical Software 2009 Correctness of Kruskal's algorithms for monotone regression with ties de Leeuw, Psychometrica, 1977 Examples -------- >>> from sklearn.datasets import make_regression >>> from sklearn.isotonic import IsotonicRegression >>> X, y = make_regression(n_samples=10, n_features=1, random_state=41) >>> iso_reg = IsotonicRegression().fit(X, y) >>> iso_reg.predict([.1, .2]) array([1.8628..., 3.7256...]) NTnanr!r"r# out_of_boundscCs||_||_||_||_dSNr/)selfr!r"r#r0rrr__init__szIsotonicRegression.__init__cCs2|jdks.|jdkr"|jddks.d}t|dS)NrzKIsotonic regression input X should be a 1d array or 2d array with 1 feature)ndimshape ValueError)r2Xmsgrrr_check_input_data_shapes"z*IsotonicRegression._check_input_data_shapecsX|jdkrtd|j|jdk}tdkr@fdd|_ntj|d|d|_d S) zBuild the f_ interp1d function.raiser.r,IThe argument ``out_of_bounds`` must be in 'nan', 'clip', 'raise'; got {0}r<rcs |jSr1)repeatr6)rrrrz-IsotonicRegression._build_f..Zlinear)Zkind bounds_errorN)r0r7formatrf_rZinterp1d)r2r8rrBrr?r_build_fs    zIsotonicRegression._build_fc sV|||d}|jdkr,t|||_n|j|_t|||jd}|dk}||||||}}}t||ffdd|||fD\}}}t |||\}}}|}t |||j |j |jd}t |t||_|_|rJtjt|ftd} tt|dd|d d t|dd|d d | dd<|| || fS||fSd S) z Build the y_ IsotonicRegression.r$autor'rcsg|] }|qSrr).0r*r-rr sz/IsotonicRegression._build_y..rrNr4)r:reshaper#rZ increasing_r r&rZlexsortr rr!r"minmaxX_min_X_max_ZonesrboolZ logical_orZ not_equal) r2r8rr Ztrim_duplicatesmaskZunique_XZunique_yZunique_sample_weightZ keep_datarrHr_build_ys<     zIsotonicRegression._build_ycCsztddd}t|fdtjtjgi|}t|fd|ji|}t|||||||\}}|||_|_ | |||S)aFit the model using X, y as training data. Parameters ---------- X : array-like of shape (n_samples,) or (n_samples, 1) Training data. .. versionchanged:: 0.24 Also accepts 2d array with 1 feature. y : array-like of shape (n_samples,) Training target. sample_weight : array-like of shape (n_samples,), default=None Weights. If set to None, all weights will be set to 1 (equal weights). Returns ------- self : object Returns an instance of self. Notes ----- X is stored for future use, as :meth:`transform` needs X to interpolate new input data. F)Z accept_sparser%r&) dictrrr(r)r&r rR X_thresholds_ y_thresholds_rE)r2r8rr Z check_paramsrrrfit)s   zIsotonicRegression.fitcCst|dr|jj}ntj}t||dd}|||d}|jdkrVt d |j|jdkrrt ||j |j }||}||j}|S)aTransform new data by linear interpolation Parameters ---------- T : array-like of shape (n_samples,) or (n_samples, 1) Data to transform. .. versionchanged:: 0.24 Also accepts 2d array with 1 feature. Returns ------- y_pred : ndarray of shape (n_samples,) The transformed data rTF)r&r%r$r;r=r,)hasattrrTr&rr(rr:rKr0r7rCr,rNrOrDZastype)r2Tr&resrrr transformXs        zIsotonicRegression.transformcCs ||S)a%Predict new data by linear interpolation. Parameters ---------- T : array-like of shape (n_samples,) or (n_samples, 1) Data to transform. Returns ------- y_pred : ndarray of shape (n_samples,) Transformed data. )rZ)r2rXrrrpredicts zIsotonicRegression.predictcst}|dd|S)z1Pickle-protocol - return state of the estimator. rDN)super __getstate__popr2state __class__rrr]s  zIsotonicRegression.__getstate__cs4t|t|dr0t|dr0||j|jdS)znPickle-protocol - set state of the estimator. We need to rebuild the interpolation function. rTrUN)r\ __setstate__rWrErTrUr_rarrrcs zIsotonicRegression.__setstate__cCs ddgiS)NZX_typesZ1darrayr)r2rrr _more_tagsszIsotonicRegression._more_tags)T)N)__name__ __module__ __qualname____doc__r r3r:rErRrVrZr[r]rcrd __classcell__rrrarrsT / /+  )ZnumpyrZscipyrZ scipy.statsrrrbaserrrZutilsrr Zutils.validationr r Z _isotonicr r __all__rrrrrrrs"  96