^j3Z dZddlZddlZddlmZddlmZddlmZm Z m Z m Z m Z m Z mZmZddlZddlmZddlmZddlZddlZddlmZmZmZmZm Z m!Z!m"Z"m#Z#m$Z$m%Z%m&Z&m'Z'm(Z(m)Z)m*Z*m+Z+m,Z,ddl-m.Z.m/Z/dd l0m1Z1m2Z2m3Z3m4Z4m5Z5m6Z6m7Z7m8Z8m9Z9m:Z:m;Z;mZ>m?Z?dd l@mAZAerdd l0mBZBgd ZCee ee eDe eEfeje?ejejjfZJee e ejejgeejejffe e ejeje ejgeejejffe e ejeje eje ejgeejejfffZKee e ejejge%fe e ejejge e%fe e ejeje ejge%fe e ejeje ejge e%fe e ejeje eje ejge%fe e ejeje eje ejge e%ffZLeeMeLe eeMeLffZNeeJe)fZOd ede ejfdZPd edejfdZQd ede ejfdZRGddZSGddZTdZUdZVdZWde ee e e fe e fdeMdeEde e fdZXde e eOde eeJeeJfde ee)ee)fde e eOfd ZYGd!d"e;ZZGd#d$es r?#_get_label_from_constructed_datasetrEs;    E F eRZZ (3)3 ( LrAcp|j}d}|!t|tjsJ||S)NzEstimators in lightgbm.sklearn should only retrieve weights from a constructed Dataset. If you're seeing this message, it's a bug in lightgbm. Please report it at https://github.com/lightgbm-org/LightGBM/issues.) get_weightr:r;r<)r6weightr>s r?$_get_weight_from_constructed_datasetrIs?    !F F >Z ;FYF ; MrAcveZdZdZdefdZdejdede ejejffdZ y) _ObjectiveFunctionWrapperz#Proxy class for objective function.funcc||_y)aJ Construct a proxy class. This class transforms objective function to match objective function with signature ``new_func(preds, dataset)`` as expected by ``lightgbm.engine.train``. Parameters ---------- func : callable Expects a callable with following signatures: ``func(y_true, y_pred)``, ``func(y_true, y_pred, weight)`` or ``func(y_true, y_pred, weight, group)`` and returns (grad, hess): y_true : numpy 1-D array of shape = [n_samples] The target values. y_pred : numpy 1-D array of shape = [n_samples] or numpy 2-D array of shape = [n_samples, n_classes] (for multi-class task) The predicted values. Predicted values are returned before any transformation, e.g. they are raw margin instead of probability of positive class for binary task. weight : numpy 1-D array of shape = [n_samples] The weight of samples. Weights should be non-negative. group : numpy 1-D array Group/query data. Only used in the learning-to-rank task. sum(group) = n_samples. For example, if you have a 100-document dataset with ``group = [10, 20, 40, 10, 10, 10]``, that means that you have 6 groups, where the first 10 records are in the first group, records 11-30 are in the second group, records 31-70 are in the third group, etc. grad : numpy 1-D array of shape = [n_samples] or numpy 2-D array of shape [n_samples, n_classes] (for multi-class task) The value of the first order derivative (gradient) of the loss with respect to the elements of y_pred for each sample point. hess : numpy 1-D array of shape = [n_samples] or numpy 2-D array of shape = [n_samples, n_classes] (for multi-class task) The value of the second order derivative (Hessian) of the loss with respect to the elements of y_pred for each sample point. .. note:: For multi-class task, y_pred is a numpy 2-D array of shape = [n_samples, n_classes], and grad and hess should be returned in the same format. NrLselfrLs r?__init__z"_ObjectiveFunctionWrapper.__init__s R rApredsr6r7c\t|}tt|jj}|dk(r|j||\}}||fSt |}|dk(r|j|||\}}||fS|dk(rt |}|j||||Std|)aCall passed function with appropriate arguments. Parameters ---------- preds : numpy 1-D array of shape = [n_samples] or numpy 2-D array of shape = [n_samples, n_classes] (for multi-class task) The predicted values. dataset : Dataset The training dataset. Returns ------- grad : numpy 1-D array of shape = [n_samples] or numpy 2-D array of shape = [n_samples, n_classes] (for multi-class task) The value of the first order derivative (gradient) of the loss with respect to the elements of preds for each sample point. hess : numpy 1-D array of shape = [n_samples] or numpy 2-D array of shape = [n_samples, n_classes] (for multi-class task) The value of the second order derivative (Hessian) of the loss with respect to the elements of preds for each sample point. zESelf-defined objective function should have 2, 3 or 4 arguments, got rElenrrL parametersrIr@ TypeError) rPrRr6labelsargcgradhessrHr=s r?__call__z"_ObjectiveFunctionWrapper.__call__s.5W=9TYY'223 19651JD$: 5g> 1965&9JD$:  197@E99VUFE: :_`d_efggrAN) __name__ __module__ __qualname____doc__#_LGBM_ScikitCustomObjectiveFunctionrQr;r<rr r_rAr?rKrKsN-)@)V&hzz&h&h rzz2::% & &hrArKc TeZdZdZdefdZdejdede e e e ffdZ y) _EvalFunctionWrapperz$Proxy class for evaluation function.rLc||_y)a Construct a proxy class. This class transforms evaluation function to match evaluation function with signature ``new_func(preds, dataset)`` as expected by ``lightgbm.engine.train``. Parameters ---------- func : callable Expects a callable with following signatures: ``func(y_true, y_pred)``, ``func(y_true, y_pred, weight)`` or ``func(y_true, y_pred, weight, group)`` and returns (metric_name, metric_value, maximize) or list of (metric_name, metric_value, maximize): y_true : numpy 1-D array of shape = [n_samples] The target values. y_pred : numpy 1-D array of shape = [n_samples] or numpy 2-D array shape = [n_samples, n_classes] (for multi-class task) The predicted values. In case of custom ``objective``, predicted values are returned before any transformation, e.g. they are raw margin instead of probability of positive class for binary task in this case. weight : numpy 1-D array of shape = [n_samples] The weight of samples. Weights should be non-negative. group : numpy 1-D array Group/query data. Only used in the learning-to-rank task. sum(group) = n_samples. For example, if you have a 100-document dataset with ``group = [10, 20, 40, 10, 10, 10]``, that means that you have 6 groups, where the first 10 records are in the first group, records 11-30 are in the second group, records 31-70 are in the third group, etc. metric_name : str Unique identifier for the metric (e.g. "custom_adjusted_mse"). metric_value : float Value of the evaluation metric. maximize : bool Are higher values better? e.g. ``True`` for AUC and ``False`` for binary error. NrNrOs r?rQz_EvalFunctionWrapper.__init__s J rArRr6r7c@t|}tt|jj}|dk(r|j||St |}|dk(r|j|||S|dk(rt |}|j||||Std|)aCall passed function with appropriate arguments. Parameters ---------- preds : numpy 1-D array of shape = [n_samples] or numpy 2-D array of shape = [n_samples, n_classes] (for multi-class task) The predicted values. dataset : Dataset The training dataset. Returns ------- metric_name : str Unique identifier for the metric (e.g. "custom_adjusted_mse"). metric_value : float Value of the evaluation metric. maximize : bool Are higher values better? e.g. ``True`` for AUC and ``False`` for binary error. rTrUrVz@Self-defined eval function should have 2, 3 or 4 arguments, got rW)rPrRr6r[r\rHr=s r?r_z_EvalFunctionWrapper.__call__s.5W=9TYY'223 1999VU+ +5g> 1999VUF3 3 197@E99VUFE: :Z[_Z`abbrAN) r`rarbrc_LGBM_ScikitCustomEvalFunctionrQr;r<rr rr r_rerAr?rgrgsM.%;%N$czz$c$c +T2N-OO P $crArga Build a gradient boosting model from the training set (X, y). Parameters ---------- X : {X_shape} Input feature matrix. .. versionadded:: 4.2.0 Support for ``pyarrow`` inputs .. versionadded:: 4.7.0 Support for ``polars`` inputs y : {y_shape} The target values (class labels in classification, real numbers in regression). .. versionadded:: 4.2.0 Support for ``pyarrow`` inputs .. versionadded:: 4.7.0 Support for ``polars`` inputs sample_weight : {sample_weight_shape} Weights of training data. Weights should be non-negative. .. versionadded:: 4.2.0 Support for ``pyarrow`` inputs .. versionadded:: 4.7.0 Support for ``polars`` inputs init_score : {init_score_shape} Init score of training data. .. versionadded:: 4.2.0 Support for ``pyarrow`` inputs .. versionadded:: 4.7.0 Support for ``polars`` inputs group : {group_shape} Group/query data. Only used in the learning-to-rank task. sum(group) = n_samples. For example, if you have a 100-document dataset with ``group = [10, 20, 40, 10, 10, 10]``, that means that you have 6 groups, where the first 10 records are in the first group, records 11-30 are in the second group, records 31-70 are in the third group, etc. .. versionadded:: 4.2.0 Support for ``pyarrow`` inputs .. versionadded:: 4.7.0 Support for ``polars`` inputs eval_set : list or None, optional (default=None) .. deprecated:: 4.7.0 A list of (X, y) tuple pairs to use as validation sets. Use ``eval_X`` and ``eval_y`` instead. eval_names : list of str, or None, optional (default=None) Unique identifiers for each evaluation dataset. Should be the same length as ``eval_set`` / ``eval_X``. eval_sample_weight : {eval_sample_weight_shape} Weights of eval data. Weights should be non-negative. eval_class_weight : list or None, optional (default=None) Class weights of eval data. eval_init_score : {eval_init_score_shape} Init score of eval data. eval_group : {eval_group_shape} Group data of eval data. eval_metric : str, callable, list or None, optional (default=None) If str, it should be a built-in evaluation metric to use. If callable, it should be a custom evaluation metric, see note below for more details. If list, it can be a list of built-in metrics, a list of custom evaluation metrics, or a mix of both. In either case, the ``metric`` from the model parameters will be evaluated and used as well. Default: 'l2' for LGBMRegressor, 'logloss' for LGBMClassifier, 'ndcg' for LGBMRanker. feature_name : list of str, or 'auto', optional (default='auto') Feature names. If 'auto' and data is pandas DataFrame, data columns names are used. categorical_feature : list of str or int, or 'auto', optional (default='auto') Categorical features. If list of int, interpreted as indices. If list of str, interpreted as feature names (need to specify ``feature_name`` as well). If 'auto' and data is pandas DataFrame, pandas unordered categorical columns are used. All values in categorical features will be cast to int32 and thus should be less than int32 max value (2147483647). Large values could be memory consuming. Consider using consecutive integers starting from zero. All negative values in categorical features will be treated as missing values. The output cannot be monotonically constrained with respect to a categorical feature. Floating point numbers in categorical features will be rounded towards 0. callbacks : list of callable, or None, optional (default=None) List of callback functions that are applied at each iteration. See Callbacks in Python API for more information. init_model : str, pathlib.Path, Booster, LGBMModel or None, optional (default=None) Filename of LightGBM model, Booster instance or LGBMModel instance used for continue training. eval_X : {X_shape}, or tuple of such inputs, or None, optional (default=None) Feature matrix or tuple thereof, e.g. ``(X_val0, X_val1)``, to use as validation sets. .. versionadded:: 4.7.0 eval_y : {y_shape}, or tuple of such inputs, or None, optional (default=None) Target values or tuple thereof, e.g. ``(y_val0, y_val1)``, to use as validation sets. .. versionadded:: 4.7.0 Returns ------- self : LGBMModel Returns self. aR Note ---- Custom eval function expects a callable with following signatures: ``func(y_true, y_pred)``, ``func(y_true, y_pred, weight)`` or ``func(y_true, y_pred, weight, group)`` and returns (metric_name, metric_value, maximize) or list of (metric_name, metric_value, maximize): y_true : numpy 1-D array of shape = [n_samples] The target values. y_pred : numpy 1-D array of shape = [n_samples] or numpy 2-D array of shape = [n_samples, n_classes] (for multi-class task) The predicted values. In case of custom ``objective``, predicted values are returned before any transformation, e.g. they are raw margin instead of probability of positive class for binary task in this case. weight : numpy 1-D array of shape = [n_samples] The weight of samples. Weights should be non-negative. group : numpy 1-D array Group/query data. Only used in the learning-to-rank task. sum(group) = n_samples. For example, if you have a 100-document dataset with ``group = [10, 20, 40, 10, 10, 10]``, that means that you have 6 groups, where the first 10 records are in the first group, records 11-30 are in the second group, records 31-70 are in the third group, etc. metric_name : str Unique identifier for the metric (e.g. "custom_adjusted_mse"). metric_value : float Value of the evaluation metric. maximize : bool Are higher values better? e.g. ``True`` for AUC and ``False`` for binary error. a_ {description} Parameters ---------- X : {X_shape} Input features matrix. raw_score : bool, optional (default=False) Whether to predict raw scores. start_iteration : int, optional (default=0) Start index of the iteration to predict. If <= 0, starts from the first iteration. num_iteration : int or None, optional (default=None) Total number of iterations used in the prediction. If None, if the best iteration exists and start_iteration <= 0, the best iteration is used; otherwise, all iterations from ``start_iteration`` are used (no limits). If <= 0, all iterations from ``start_iteration`` are used (no limits). pred_leaf : bool, optional (default=False) Whether to predict leaf index. pred_contrib : bool, optional (default=False) Whether to predict feature contributions. .. note:: If you want to get more explanations for your model's predictions using SHAP values, like SHAP interaction values, you can install the shap package (https://github.com/slundberg/shap). Note that unlike the shap package, with ``pred_contrib`` we return a matrix with an extra column, where the last column is the expected value. validate_features : bool, optional (default=False) If True, ensure that the features used to predict match the ones used to train. Used only if data is pandas DataFrame. **kwargs Other parameters for the prediction. Returns ------- {output_name} : {predicted_result_shape} The predicted values. X_leaves : {X_leaves_shape} If ``pred_leaf=True``, the predicted leaf of every tree for each sample. X_SHAP_values : {X_SHAP_values_shape} If ``pred_contrib=True``, the feature contributions for each sample. collectionnameic|yt|trt||kDr||Syt|tr|j |dSt |d)z?Try to extract the ith element of one of the ``eval_*`` inputs.Nz should be dict or list)r:listrXdictgetrZrkrlrms r?_extract_evaluation_meta_datarss` J % z?Q a=  J %~~a&&4& 7899rAeval_seteval_Xeval_yc|Bd}tj|td|| tdt |t r|gS|S|du|duk7r td||t |t t |t k7r tdt |t r2t |t r"t |t |k7r tdt |t r)t |t rtt||d }|S||fg}|S) z>Validate eval args. Returns ------- eval_set NzIThe argument 'eval_set' is deprecated, use 'eval_X' and 'eval_y' instead.rT)category stacklevelzASpecify either 'eval_set' or 'eval_X' and 'eval_y', but not both.z9You must specify eval_X and eval_y, not just one of them.zKIf eval_X is a tuple, y_val must be a tuple of same length, and vice versa.Tstrict) warningswarnr ValueErrorr:tuplerXrozip)rtrurvmsgs r?_validate_eval_set_XyrsY c$:qI  !3`a a h &: O$FdN+TUUF. fe $ 65(A Ajk k fe $FE)B6{c&k) !noo fe $FE)BCt<=H O ()H OrAc*\eZdZdZddddddddd d d d d d d d dddddededededededeeee fdeee efdedededededededed eeee jje jjfd!eed"ed#ef(d$Zd%e eeffd&Zed'd(d)e eefd%d(fd*Zd%ed(ffd+ Zd%efd,Zdod-ed%e eefffd. Zd/ed%dfd0Zd1ed%e eeffd2Zd!eed%efd3Z dpddd4d5ed6ed7eed8eed9eed:ee e!d;ee edee ed?ee ed@ee"dAe#dBe$dCee e%dDeeee&e'dfdEeeee(efdFeeee(efd%df&dGZ)e*jWdHdIdJdKdLdMdNdOPdQze,ze)_ dqd5edRedSedTeedUedVedWed#ed%e-fdXZ.e/jWdYdZd[d\d]d^_e._e0d%efd`Z1e0d%efdaZ2e2jfdbed%dfdcZ2e0d%e4fddZ5e0d%efdeZ6e0d%eee ffdfZ7e0d%efdgZ8e0d%efdhZ9e0d%e'fdiZ:e0d%e;fdjZe0d%e efdlZ?e0d%e jzfdmZ@e@jdrdnZ@xZBS)sr3z4Implementation of the scikit-learn API for LightGBM.gbdt皙?d@ NMbP??rsplit boosting_type num_leaves max_depth learning_rate n_estimatorssubsample_for_bin objective class_weightmin_split_gainmin_child_weightmin_child_samples subsamplesubsample_freqcolsample_bytree reg_alpha reg_lambda random_staten_jobsimportance_typerrrrrrrrrrrrrrrrrrrkwargsc ts td||_||_||_||_||_||_||_| |_ | |_ | |_ | |_ | |_ ||_||_||_||_||_||_d|_i|_i|_d|_i|_||_||_d|_d|_d|_d|_d|_d|_ d|_!|jDdi|y)aConstruct a gradient boosting model. Parameters ---------- boosting_type : str, optional (default='gbdt') 'gbdt', traditional Gradient Boosting Decision Tree. 'dart', Dropouts meet Multiple Additive Regression Trees. 'rf', Random Forest. num_leaves : int, optional (default=31) Maximum tree leaves for base learners. max_depth : int, optional (default=-1) Maximum tree depth for base learners, <=0 means no limit. If setting this to a positive value, consider also changing ``num_leaves`` to ``<= 2^max_depth``. learning_rate : float, optional (default=0.1) Boosting learning rate. You can use ``callbacks`` parameter of ``fit`` method to shrink/adapt learning rate in training using ``reset_parameter`` callback. Note, that this will ignore the ``learning_rate`` argument in training. n_estimators : int, optional (default=100) Number of boosted trees to fit. subsample_for_bin : int, optional (default=200000) Number of samples for constructing bins. objective : str, callable or None, optional (default=None) Specify the learning task and the corresponding learning objective or a custom objective function to be used (see note below). Default: 'regression' for LGBMRegressor, 'binary' or 'multiclass' for LGBMClassifier, 'lambdarank' for LGBMRanker. class_weight : dict, 'balanced' or None, optional (default=None) Weights associated with classes in the form ``{class_label: weight}``. Use this parameter only for multi-class classification task; for binary classification task you may use ``is_unbalance`` or ``scale_pos_weight`` parameters. Note, that the usage of all these parameters will result in poor estimates of the individual class probabilities. You may want to consider performing probability calibration (https://scikit-learn.org/stable/modules/calibration.html) of your model. The 'balanced' mode uses the values of y to automatically adjust weights inversely proportional to class frequencies in the input data as ``n_samples / (n_classes * np.bincount(y))``. If None, all classes are supposed to have weight one. Note, that these weights will be multiplied with ``sample_weight`` (passed through the ``fit`` method) if ``sample_weight`` is specified. min_split_gain : float, optional (default=0.) Minimum loss reduction required to make a further partition on a leaf node of the tree. min_child_weight : float, optional (default=1e-3) Minimum sum of instance weight (Hessian) needed in a child (leaf). min_child_samples : int, optional (default=20) Minimum number of data needed in a child (leaf). subsample : float, optional (default=1.) Subsample ratio of the training instance. subsample_freq : int, optional (default=0) Frequency of subsample, <=0 means no enable. colsample_bytree : float, optional (default=1.) Subsample ratio of columns when constructing each tree. reg_alpha : float, optional (default=0.) L1 regularization term on weights. reg_lambda : float, optional (default=0.) L2 regularization term on weights. random_state : int, RandomState object or None, optional (default=None) Random number seed. If int, this number is used to seed the C++ code. If RandomState or Generator object (numpy), a random integer is picked based on its state to seed the C++ code. If None, default seeds in C++ code are used. n_jobs : int or None, optional (default=None) Number of parallel threads to use for training (can be changed at prediction time by passing it as an extra keyword argument). For better performance, it is recommended to set this to the number of physical cores in the CPU. Negative integers are interpreted as following joblib's formula (n_cpus + 1 + n_jobs), just like scikit-learn (so e.g. -1 means using all threads). A value of zero corresponds the default number of threads configured for OpenMP in the system. A value of ``None`` (the default) corresponds to using the number of physical cores in the system (its correct detection requires either the ``joblib`` or the ``psutil`` util libraries to be installed). .. versionchanged:: 4.0.0 importance_type : str, optional (default='split') The type of feature importance to be filled into ``feature_importances_``. If 'split', result contains numbers of times the feature is used in a model. If 'gain', result contains total gains of splits which use the feature. **kwargs Other parameters for the model. Check http://lightgbm.readthedocs.io/en/latest/Parameters.html for more parameters. .. warning:: \*\*kwargs is not supported in sklearn, it may cause unexpected issues. Note ---- A custom objective function can be provided for the ``objective`` parameter. In this case, it should have the signature ``objective(y_true, y_pred) -> grad, hess``, ``objective(y_true, y_pred, weight) -> grad, hess`` or ``objective(y_true, y_pred, weight, group) -> grad, hess``: y_true : numpy 1-D array of shape = [n_samples] The target values. y_pred : numpy 1-D array of shape = [n_samples] or numpy 2-D array of shape = [n_samples, n_classes] (for multi-class task) The predicted values. Predicted values are returned before any transformation, e.g. they are raw margin instead of probability of positive class for binary task. weight : numpy 1-D array of shape = [n_samples] The weight of samples. Weights should be non-negative. group : numpy 1-D array Group/query data. Only used in the learning-to-rank task. sum(group) = n_samples. For example, if you have a 100-document dataset with ``group = [10, 20, 40, 10, 10, 10]``, that means that you have 6 groups, where the first 10 records are in the first group, records 11-30 are in the second group, records 31-70 are in the third group, etc. grad : numpy 1-D array of shape = [n_samples] or numpy 2-D array of shape = [n_samples, n_classes] (for multi-class task) The value of the first order derivative (gradient) of the loss with respect to the elements of y_pred for each sample point. hess : numpy 1-D array of shape = [n_samples] or numpy 2-D array of shape = [n_samples, n_classes] (for multi-class task) The value of the second order derivative (Hessian) of the loss with respect to the elements of y_pred for each sample point. For multi-class task, y_pred is a numpy 2-D array of shape = [n_samples, n_classes], and grad and hess should be returned in the same format. zyscikit-learn is required for lightgbm.sklearn. You must install scikit-learn and restart your session to use this module.NrFre)#r"rrrrrrrrrrrrrrrrrrr_Booster _evals_result _best_score_best_iteration _other_params _objectiver _class_weight _class_map_fitted_with_feature_names _n_features_n_features_in_classes _n_classes set_params)rPrrrrrrrrrrrrrrrrrrrrs r?rQzLGBMModel.__init__Bs"\!]  +"$"*(!2, 0!2", 0"$( .+/ .079$&-/#(9=4805' "#%.2 !!&!rAr7cd}dd|||d}tjd^}}}t|t|fdkrd|d<d gd |d S) NaIn LightGBM, setting a sample's weight to 0 can produce a different result than omitting the sample. Such samples intentionally still affect count-based measures like 'min_data_in_leaf' (https://github.com/lightgbm-org/LightGBM/issues/5626#issuecomment-1712706678) and the estimated distribution of features for Dataset construction (see https://github.com/lightgbm-org/LightGBM/issues/5553).zscikit-learn incorrectly asserts that private attributes cannot be set in __init__: (see https://github.com/lightgbm-org/LightGBM/issues/2628)aSBeginning in scikit-learn 1.9, by default estimators are expected to reject sample weight arrays that are all-0. LightGBM intentionally accepts such arrays. LightGBM supports some operations where training on an all-0-weight input could make sense, like batch updates with training continuation or manual model creation with forced splits.)check_no_attributes_set_in_init#check_all_zero_sample_weights_errorcheck_sample_weight_equivalence-check_sample_weight_equivalence_on_dense_data.check_sample_weight_equivalence_on_sparse_data.)r rTzdecision_function() returns raw margins while predict_proba() applies sigmoid in C++ independently, causing different tie structures after rounding. scikit-learn >= 1.2 relaxed this check to accept monotonically consistent scores. check_decision_proba_consistencyT)2darraysparse1dlabels) allow_nanX_types _xfail_checks)r.rint)rPcheck_sample_weight_str xfail_checks sklearn_major sklearn_minor_s r? _more_tagszLGBMModel._more_tagss o M m 0G=T>U $,<+A+A#+F( }q  M 2 3f <d ; < 8)  rAtagsr1 tags_dictc|d|j_d|dv|j_d|dv|j_|S)aSUpdate ``sklearn.utils.Tags`` inherited from ``scikit-learn`` base classes. ``scikit-learn`` 1.6 introduced a dataclass-based interface for estimator tags. ref: https://github.com/scikit-learn/scikit-learn/pull/29677 This method handles updating that instance based on the value in ``self._more_tags()``. rrrr) input_tagsrr target_tags one_d_labelsrrs r?_update_sklearn_tags_from_dictz(LGBMModel._update_sklearn_tags_from_dict%sH%.k$:!!)Yy-A!A(2i 6J(J% rActtdsdt}t||j t ||jS)N__sklearn_tags__zW__sklearn_tags__() should not be called when using scikit-learn<1.6. Detected version: r)hasattrr+r.AttributeErrorrsuperrr)rPerr_msg __class__s r?rzLGBMModel.__sklearn_tags__7sa~'9:%%5$68 !) )22)+oo'3  rAct|ddS)Nfitted_F)getattrrPs r?__sklearn_is_fitted__zLGBMModel.__sklearn_is_fitted__HstY..rAdeepct||}tj|}tt |j j dddr|j j d|_n|j j d|_|j|j j|||j|j|S)agGet parameters for this estimator. Parameters ---------- deep : bool, optional (default=True) If True, will return the parameters for this estimator and contained subobjects that are estimators. Returns ------- params : dict Parameter names mapped to their values. )rr get_paramsNr ) rrcopycallablerr __bases__updater)rPrparamscprs r?rzLGBMModel.get_paramsKsB##. YYt_ GBLL2215|TJ K<<11!4BL<<11!4BL bll--b$78 d(() rArc |jD]@\}}t|||t|d|rt|d||||j|<B|S)zSet the parameters of this estimator. Parameters ---------- **params Parameter names with their new values. Returns ------- self : object Returns self. r)itemssetattrrr)rPrkeyvalues r?rzLGBMModel.set_paramsysb!,,. ,JC D#u %tqY'#i/&+D  s #  ,  rAstagecJ|dvsJ|j}|jddtjdD]4}||vs|j|}t d|d|dk(s.||_6|dk(rv|j jt |trd|_nRt |tr|jdkDrd |_n+d |_n#t |trd |_n td t|j r$|dk(rt|j |d<nd |d<n|j |d<|jdd|jdd|jddt |dtjj rC|dj#tj$tj&j(|d<nrt |dtjj*rKt-|dj/tj$tj&j(|d<|jdkDr;tjdD]}|j|d|j|d<t1|drP|j2}tjdD]'}||vst d|d|j|})||d<t |j t4r |j nd}|It |trd}n6t |tr|jdkDrdnd}nt |trd}t7d||}|dk(r.t7d||j8}|j;|d|d<|S)apProcess the parameters of this estimator based on its type, parameter aliases, etc. Parameters ---------- stage : str Name of the stage (can be ``fit`` or ``predict``) this method is called from. Returns ------- processed_params : dict Processed parameter names mapped to their values. >fitpredictrNzFound 'z8' in params. Will use it instead of 'objective' argumentr regressionrT multiclassbinary lambdarankzUnknown LGBMModel type.Nonerrrr num_class_eval_ateval_atz6' in params. Will use it instead of 'eval_at' argumentl2 multi_loglossbinary_loglossndcgmetric num_threads)rpoprrqrrr:r5r2rr4r~rrKr;random RandomStaterandintiinfoint32max Generatorrintegersrrstrrr_process_n_jobs)rPrraliasobjroriginal_metrics r?_process_paramszLGBMModel._process_paramss&****" ;%#'' 4 *Ejj'wug-efgE>&)DO  * E>&dM2&2DOn5**6*2j1&2DO$%>?? DOO $~&?&P{#&,{#"&//F;  $d+ >4( >4( f^,bii.C.C D%+N%;%C%CBHHRXXDVDZDZ%[F> " ~. 0C0C D%()?)H)HRTRZRZI[I_I_)`%aF> " ??Q '++K8 ( 5$' ("&//F;  4 $mmG'++I6 0F? 75'1g!hi$jj/G 0!(F9 .8-M$//SW  "$ ."&D.159__q5H/N^D*-"(%XvG E>( LF$($8$8 9N$OF= ! rAch|td}|S|dkrttddz|zd}|S)aConvert special values of n_jobs to their actual values according to the formulas that apply. Parameters ---------- n_jobs : int or None The original value of n_jobs, potentially having special values such as 'None' or negative integers. Returns ------- n_jobs : int The value of n_jobs with special values converted to actual number of threads. T)only_physical_coresrFr )r)r)rPrs r?rzLGBMModel._process_n_jobssD >"t0|$jAD%&cic]\}%}&|j>|%|&}$}%}&t+|$|!d}'|#t-|#d k(r|'}#nt/j0|#|'}#t;| d| }(t;| d| })t3|!d |!d|#|)|(d|}"|jC|"t|tDr |jF}|g}ntj|}i}*|jCtI|*tK|||jL||||||_'|jNjQ|_)|jT|_+|*|_,|jNjZ|_.|jNj^|_0d|_1|jNje~~|Scc}wcc}wcc}wcc}wcc}&}%w)4Docstring is set after definition, using a template.rrNrr TFrTreset accept_sparseensure_all_finiteensure_min_samples)allow_all_zero_weightsr)datarDrHr=rrrr)rtrurvzLength of eval_group (z#) not equal to length of eval_set ()rrrrrrauto)r%rDrHr=rrr)r train_setnum_boost_round valid_sets valid_namesfevalrr)3r r:rordeepcopyrrgrtyper/shapen_features_in_nwdis_into_dataframenw from_nativer-r!r&rrr(rXr;multiplyrrr~ enumeratersrprrappendr3booster_r r0rr num_featurer_has_non_default_feature_namesrrbest_iterationr best_scorerr free_dataset)+rPrrrrr=rtrrrrrrrrrrrurvreval_metric_listfeval_metrics_callablemeval_metrics_builtiner_X_yclass_sample_weightr(r*rm valid_data valid_set valid_weightvalid_class_weightkvvalid_class_sample_weightvalid_init_score valid_group evals_results+ r?rz LGBMModel.fits.%%E%2  !   T *#}}[9  $ k :; CS bQV^_`Va!5a!8 b b+;Raz!S?QRR2#00         [[]  ,E   ud + , f%-]NDKKX(,(<(<^M=Z([}%$t}}$$   +'%/    rAz+Return the predicted value for each sample.mnumpy array, pandas DataFrame, scipy.sparse, list of lists of int or float of shape = [n_samples, n_features]predicted_resultCarray-like of shape = [n_samples] or shape = [n_samples, n_classes]Varray-like of shape = [n_samples, n_trees] or shape = [n_samples, n_trees * n_classes]array-like of shape = [n_samples, n_features + 1] or shape = [n_samples, (n_features + 1) * n_classes] or list with n_classes length of such objects descriptionrQ output_namepredicted_result_shapeX_leaves_shapeX_SHAP_values_shapecP|js td|jS)3:obj:`int`: The number of features of fitted model.z1No n_features found. Need to call fit beforehand.)rr#rrs r? n_features_zLGBMModel.n_features_ ())+$%XY YrAcP|js td|jS)roz4No n_features_in found. Need to call fit beforehand.)rr#rrs r?r0zLGBMModel.n_features_in_s())+$%[\ \"""rArc||_y)aSet number of features found in passed-in dataset. Starting with ``scikit-learn`` 1.6, ``scikit-learn`` expects to be able to directly set this property in functions like ``validate_data()``. .. note:: Do not call ``estimator.n_features_in_ = some_int`` or anything else that invokes this method. It is only here for compatibility with ``scikit-learn`` validation functions used internally in ``lightgbm``. N)r)rPrs r?r0zLGBMModel.n_features_in_s $rAcP|js td|jS)z,:obj:`dict`: The best score of fitted model.z1No best_score found. Need to call fit beforehand.)rr#rrs r? best_score_zLGBMModel.best_score_'rqrAcP|js td|jS)zc:obj:`int`: The best iteration of fitted model if ``early_stopping()`` callback has been specified.zRNo best_iteration found. Need to call fit with early_stopping callback beforehand.)rr#rrs r?best_iteration_zLGBMModel.best_iteration_.s.))+$d ###rAcP|js td|jS)zT:obj:`str` or :obj:`callable`: The concrete objective used while fitting this model.z0No objective found. Need to call fit beforehand.)rr#rrs r? objective_zLGBMModel.objective_7s&))+$%WX XrAcl|js td|jjS):obj:`int`: True number of boosting iterations performed. This might be less than parameter ``n_estimators`` if early stopping was enabled or if boosting stopped early due to limits on complexity like ``min_gain_to_split``. .. versionadded:: 4.0.0 z3No n_estimators found. Need to call fit beforehand.rr#rcurrent_iterationrs r? n_estimators_zLGBMModel.n_estimators_>s/))+$%Z[ [}}..00rAcl|js td|jjS)r{z-No n_iter found. Need to call fit beforehand.r|rs r?n_iter_zLGBMModel.n_iter_Ks/))+$%TU U}}..00rAcP|js td|jS)z.Booster: The underlying Booster of this model.z.No booster found. Need to call fit beforehand.)rr#rrs r?r8zLGBMModel.booster_X&))+$%UV V}}rAcP|js td|jS)zK:obj:`dict`: The evaluation results if validation sets have been specified.z $"!'OS37 #"&!#"%Y] $&+t"t" t"  t"  t"t"t"E#'J"JKLt"uT3Y/0t"t" t"t"t"t"  !t""#t"$%t"&uS"))*?*?ATAT%TUV't"( )t"*+t",-t"t' DcN' RS>  " (?"; "/t/,t,tCH~,\3;(TSTT#s(^Tlhsm05948+/9=*.?C37?C6:<@7=EK.2GK#{&Z^KO){ !{ { 01 { 01 { ( {4 456{T#Y'{%T*:%;<{$DK0{"$':";<{T/23{89{5{C{ DN+!{"U3g{#BCD#{&5u=S7TTUV'{(o0FFGH){* +{| !!eW!pbR&B"{q "    * *K$  '+""'< !< <  <  } <  < <  < <  !< |-33A&dos 4GO S  ###  $C $D $ $ 7  $$$E#'J"JK  1s 1 1 1 1 1' """  Vbjj V V ,tCy , ,,2::,,&  rAr3c+eZdZdZddddddddd d d d d d d d dddddededededededeeee fdeee efdedededededededed eeee jje jjfd!eed"ed#ed$df*fd%Zej je_d$e eeffd&ZdAfd' Z dBddd(d)ed*ed+eed,eed-eeed.eeed/eeed0eeed1eed2ed3ed4eeed5eeeeeefd6eeee efd7eeee efd$df fd8Z!ejBjjEd9d:Z#e#de#jId;e#e#jId<dzZ#e#de#jId=e#e#jId>dzZ#e#de#jId?e#e#jId@dze!_xZ%S)Cr5zLightGBM regressor.rrrrrrNrrrrrrrrrrrrrrrrrrrrrrrrrrrr7c t|did|d|d|d|d|d|d|d|d | d | d | d | d | d|d|d|d|d|d||yNrrrrrrrrrrrrrrrrrrrrerrQrPrrrrrrrrrrrrrrrrrrrrrs r?rQzLGBMRegressor.__init__0  ' !   (  &  0    & * . 0   * .   "! "&# $% &,) rActttddrtj|}ni}|j t j||SNr)rrr,rrr3rPrs r?rzLGBMRegressor._more_tagssE G. dC D%006DD I((./ rAc t|S)N)rr)rPrs r?rzLGBMRegressor.__sklearn_tags__sw'))rAr rrrrrtrrrrrrrrrurvcDt|||||||||||| | | | | |S)*Docstring is inherited from the LGBMModel.) rrrtrurvrrrrrrrr)rr)rPrrrrrtrrrrrrrrrurvrs r?rzLGBMRegressor.fitsJ(  '!!1+#% 3!  " rAself : LGBMModelzself : LGBMRegressorgroup : eval_set :eval_class_weight :eval_init_score : eval_group : eval_metric :r7r1) NNNNNNNr'r'NN)&r`rarbrcrrrr r rdrr;rrrrrQr3rrrrrrr rrrrrrrr rreplace _base_docfindrrs@r?r5r5s$"!'OS37 #"&!#"%Y] $&+- -  -  -  - - - E#'J"JKL- uT3Y/0- -  - - - -  !- "#- $%- &uS"))*?*?ATAT%TUV'- ( )- *+- ,-- . /- ^!))11HDcN*59489=*.?C?C<@7=EK.2EI% Z^KO#% !% % 01 % 01 % 4 456 %T#Y'%%T*:%;<%"$':";<%89%5%C%DN+%U3gy#@AB% 5u=S7TTUV!%"o0FFGH#%$ %%N %%--.@BXYI-INN9-. INN<02 3 4AINN+@ABYy~~^qOrOtEuuIe#e#jId?dze!_ dWd)edAe%dBedCeedDe%dEe%dFe%d#ed$e&fdGZ'ejNje'_ dWd)edAe%dBedCeedDe%dEe%dFe%d#ed$e&ffdH Z(e)jUdIdJdKdLdMdNOe(_d dd@dPd)edBedCeedFe%d#ed$e&f fdQZ+e,d$e jZfdRZ.e,d$efdSZ/e,d$e%fdTZ0xZ1S)Xr2zLightGBM classifier.rrrrrrNrrrrrrrrrrrrrrrrrrrrrrrrrrrr7c t|did|d|d|d|d|d|d|d|d | d | d | d | d | d|d|d|d|d|d||yrrrs r?rQzLGBMClassifier.__init__!rrActttddrtj|}ni}|j t j||Sr)rrr'rrr3rs r?rzLGBMClassifier._more_tagsRsE G/tD E&11$7DD I((./ rAclt|}|"d|j_d|j_|S)NTF)rrclassifier_tags multi_class multi_label)rPrrs r?rzLGBMClassifier.__sklearn_tags__\s7w')  /3D ,/4D , rAr rrrrrtrrrrrrrrrrurvct|t|tj||_|jj |}t t|jj|jj |jjd|_ t|jt r@|jjDcic]\}}|j||c}}|_ |jj|_t|j|_|j"d|_t'| st| t(r| }nt| t*r| g}ng}|j,r(t/|D]\}}|dvrd||<|dvsd||<n't/|D]\}}|dvrd ||<|d vsd ||<|} d}|gt|t0r|g}g}|D]M\}}||ur||ur|j3||f"|j3||jj |fOt4| ||||||||||| | | | | | |Scc}}w) rTrzN>loglossrr>error binary_error multi_error>rrr>rrr)rrrtrrurvrrrrrrrr)r$r%r*r_le transformrprclasses_rr:rrrrrXrrrrror_LGBMClassifier__is_multiclassr6rr7r)rPrrrrrtrrrrrrrrrrurvrErKrLr>indexrr*valid_xvalid_yrs r?rzLGBMClassifier.fitcs* Q'*$&**1- XX   "s488#4#4dhh6H6HIZIZ6[dhij d'' .DHDUDUD[D[D]!^DAq$//!"4a"7!^D )) dmm, >> !"DO $+t,#. K-$/= #% ##%./?%@@ME6!>>2A(/#<<2?(/ @ &//?%@AME6!==2B(/#;;2@(/ A +K<@  (E*$:J$, N ars r?rzLGBMClassifier.predict_probas  +'%/     DOO $i9 0  M  ! !Y)|MN fbjj1 <9 <199cFlF34>>@ @rAz@Return the predicted probability for each class for each sample.rcpredicted_probabilityrerfrgrh)rZr[r^c .t|d|d|||d|S)aReturn the raw margin score for each sample. .. versionadded:: 4.7.0 Parameters ---------- X : numpy array, pandas DataFrame, scipy.sparse, list of lists of int or float of shape = [n_samples, n_features] Input features matrix. start_iteration : int, optional (default=0) Start index of the iteration to predict. If <= 0, starts from the first iteration. num_iteration : int or None, optional (default=None) Total number of iterations used in the prediction. If None, if the best iteration exists and start_iteration <= 0, the best iteration is used; otherwise, all iterations from ``start_iteration`` are used (no limits). If <= 0, all iterations from ``start_iteration`` are used (no limits). validate_features : bool, optional (default=False) If True, ensure that the features used to predict match the ones used to train. Used only if data is pandas DataFrame. **kwargs Other parameters forwarded to ``predict()``. Returns ------- raw_score : array-like of shape = [n_samples] or shape = [n_samples, n_classes] The predicted values. T)rrYrZr[r^re)rr)rPrrZr[r^rrs r?decision_functionz LGBMClassifier.decision_functions5Hw +'/     rAcP|js td|jS)z;:obj:`array` of shape = [n_classes]: The class label array..No classes found. Need to call fit beforehand.)rr#rrs r?rzLGBMClassifier.classes_>rrAcP|js td|jS)z":obj:`int`: The number of classes.r)rr#rrs r? n_classes_zLGBMClassifier.n_classes_Es&))+$%UV VrAc|jdkDxs.t|jtxr|jtvS)zI:obj:`bool`: Indicator of whether the classifier is used for multiclass.rT)rr:rrrrs r?__is_multiclasszLGBMClassifier.__is_multiclassLs2"vz$//3'G'uDOO_uLuvrAr) NNNNNNNNr'r'NNr)2r`rarbrcrrrr r rdrr;rrrrrQr3rrrrrrr rrrrrrrr rrrrrrrrrrrrr<rrrrrs@r?r2r2st$"!'OS37 #"&!#"%Y] $&+- -  -  -  - - - E#'J"JKL- uT3Y/0- -  - - - -  !- "#- $%- &uS"))*?*?ATAT%TUV'- ( )- *+- ,-- . /- ^!))11HDcN59489=*.?C37?C<@7=EK.2EIW"Z^KO%W !W W 01 W 01 W 4 456 WT#Y'W%T*:%;<W$DK0W"$':";<W89W5WCWDN+WU3gy#@ABW"5u=S7TTUV#W$o0FFGH%W& 'Wr %%--.@BYZI-INN9-. INN<02 3 4dze _e jZ"e"jId?\Z%Z&Z'e%d@e&e'e _xZ(S)Br4a!LightGBM ranker. .. warning:: scikit-learn doesn't support ranking applications yet, therefore this class is not really compatible with the sklearn ecosystem. Please use this class mainly for training and applying ranking models in common sklearnish way. rrrrrrNrrrrrrrrrrrrrrrrrrrrrrrrrrrr7c t|did|d|d|d|d|d|d|d|d | d | d | d | d | d|d|d|d|d|d||yrrrs r?rQzLGBMRanker.__init__^rrAr rrrrr=rtrrrrrr.rrrrrurvc| td| ||| td| |_t| ||||||||||| | | | ||||S)rz!Should set group for ranking taskzLeval_group cannot be None if any of eval_set, eval_X, or eval_y are provided)rrr=rtrrurvrrrrrrrr)r~rrr)rPrrrrr=rtrrrrrrrrrrrurvrs r?rzLGBMRanker.fits0 =@A A  8#76;MQWQckl l   '!!1+!#% 3!#  & rArzself : LGBMRankerrrzfeature_name :zeval_at : list or tuple of int, optional (default=(1, 2, 3, 4, 5)) The evaluation positions of the specified metric. )NNNNNNNNN)r rTrUrVr'r'NN))r`rarbrcrrrr r rdrr;rrrrrQr3rrrrrr rrr rrrrrrrrr partition_before_feature_name _feature_name_after_feature_namerrs@r?r4r4Rs$"!'OS37 #"&!#"%Y] $&+- -  -  -  - - - E#'J"JKL- uT3Y/0- -  - - - -  !- "#- $%- &uS"))*?*?ATAT%TUV'- ( )- *+- ,-- . /- ^!))11H 5948+/9=*.?C?C6:<@5D7=EK.2EI#2&Z^KO)2 !2 2 01 2 01 2 ( 24 4562T#Y'2%T*:%;<2"$':";<2T/232892tCy%S/12252C2 DN+!2"U3gy#@AB#2&5u=S7TTUV'2(o0FFGH)2* +2h %%--.@BUVI9INN#89: INN#679 : ;K I?H?R?RSc?d<-)<+,-O'(,CKKrAr4)^rcrr|inspectrpathlibrtypingrrrrr r r r narwhalsr3narwhals.dependencies dependenciesr1narwhals.typingnwtnumpyr; scipy.sparsescipybasicrrrrrrrrrrrrrrrrrcallbackrr compatr!r"r#r$r%r&r'r(r)r*r+r,r-r.r/enginer0r1__all__rrr< IntoDataFramerspmatrixrrdrjrrrr@rErIrKrgrrrrsrr3r5r2r4rerAr?rs2 SSS#&9"% tE{DI% &'JJ LL ', "** rzz* bjj"**$% '  "** rzz8BJJ+?@ bjj"**$% '  "** rzz8BJJ+?"**AUV bjj"**$% ' '#""' "** rzz*$ & "** rzz* )* ,  "** rzz8BJJ+?@$ & "** rzz8BJJ+?@ )* ,  "** rzz8BJJ+?"**AUV$ & "** rzz8BJJ+?"**AUV )* ,-":#"s22 346 3_DEXbjj=QRZZ'hrzz>RThThnNcNchkZ#>,^:tCH~tCy89:: : : c] :0!t012! U159O3PPQ R! U?E/,BBC D ! d'() !Hg  g  Tpm& pmfsw()swl z,z,rA