mlquantify 0.1.7__py3-none-any.whl → 0.1.9__py3-none-any.whl

mlquantify/methods/meta.py

@@ -1,472 +0,0 @@
-import numpy as np
-import pandas as pd
-from copy import deepcopy
-from tqdm import tqdm
-from sklearn.linear_model import LogisticRegression
-from sklearn.model_selection import GridSearchCV, cross_val_predict
-from ..evaluation import measures
-from ..base import Quantifier
-from ..utils.method import getHist, hellinger
-from ..utils.general import make_prevs, normalize_prevalence, parallel, get_indexes_with_prevalence
-
-class Ensemble(Quantifier):
-    """Ensemble of Quantification Models.
-    
-    This class implements an ensemble of quantification methods, 
-    allowing parallel processing for evaluation. The ensemble 
-    method is based on the articles by Pérez-Gállego et al. (2017, 2019).
-    
-    This approach of Ensemble is made of taking multiple
-    samples varying class proportions on each, and for the 
-    predictions, it takes the k models which as the minimum
-    seletion metric
-    
-    Attributes
-    ----------
-    base_quantifier : Quantifier
-        The base quantifier model to be used in the ensemble.
-    size : int
-        The number of samples to be generated for the ensemble.
-    min_prop : float
-        The minimum proportion of each class in the generated samples.
-    selection_metric : str
-        The metric used for selecting the best models in the ensemble.
-        Valid options are 'all', 'ptr', and 'ds'.
-        - all -> return all the predictions
-        - ptr -> computes the selected error measure
-        - ds -> computes the hellinger distance of the train and test
-     distributions for each model
-    p_metric : float
-        The proportion of models to be selected based on the selection metric.
-    return_type : str
-        The type of aggregation to be used for the final prediction.
-        Valid options are 'mean' and 'median'.
-    max_sample_size : int or None
-        The maximum size of the samples to be generated. If None, the entire dataset is used.
-    max_trials : int
-        The maximum number of trials to generate valid samples.
-    n_jobs : int
-        The number of parallel jobs to run.
-    verbose : bool
-        If True, prints progress messages during fitting and prediction.
-        
-    See Also
-    --------
-    joblib.Parallel : Parallel processing utility for Python.
-    
-    Parameters
-    ----------
-    quantifier : Quantifier
-        The base quantifier model to be used in the ensemble.
-    size : int, optional (default=50)
-        The number of samples to be generated for the ensemble.
-    min_prop : float, optional (default=0.1)
-        The minimum proportion of each class in the generated samples.
-    selection_metric : str, optional (default='all')
-        The metric used for selecting the best models in the ensemble.
-        Valid options are 'all', 'ptr', and 'ds'.
-    p_metric : float, optional (default=0.25)
-        The proportion of models to be selected based on the selection metric.
-    return_type : str, optional (default='mean')
-        The type of aggregation to be used for the final prediction.
-        Valid options are 'mean' and 'median'.
-    max_sample_size : int or None, optional (default=None)
-        The maximum size of the samples to be generated. If None, the entire dataset is used.
-    max_trials : int, optional (default=100)
-        The maximum number of trials to generate valid samples.
-    n_jobs : int, optional (default=1)
-        The number of parallel jobs to run.
-    verbose : bool, optional (default=False)
-        If True, prints progress messages during fitting and prediction.
-        
-    References
-    ----------
-    .. [1] PÉREZ-GÁLLEGO, Pablo; QUEVEDO, José Ramón; DEL COZ, Juan José. Using ensembles for problems with characterizable changes in data distribution: A case study on quantification. Information Fusion, v. 34, p. 87-100, 2017. Avaliable at https://www.sciencedirect.com/science/article/abs/pii/S1566253516300628?casa_token=XblH-3kwhf4AAAAA:oxNRiCdHZQQa1C8BCJM5PBnFrd26p8-9SSBdm8Luf1Dm35w88w0NdpvoCf1RxBBqtshjyAhNpsDd
-    .. [2] PÉREZ-GÁLLEGO, Pablo et al. Dynamic ensemble selection for quantification tasks. Information Fusion, v. 45, p. 1-15, 2019. Avaliable at https://www.sciencedirect.com/science/article/abs/pii/S1566253517303652?casa_token=jWmc592j5uMAAAAA:2YNeZGAGD0NJEMkcO-YBr7Ak-Ik7njLEcG8SKdowLdpbJ0mwPjYKKiqvQ-C3qICG8yU0m4xUZ3Yv        
-
-    Examples
-    --------
-    >>> from mlquantify.methods import FM, Ensemble
-    >>> from mlquantify.utils.general import get_real_prev
-    >>> from sklearn.ensemble import RandomForestClassifier
-    >>> from sklearn.datasets import load_breast_cancer
-    >>> from sklearn.model_selection import train_test_split
-    >>> 
-    >>> features, target = load_breast_cancer(return_X_y=True)
-    >>> 
-    >>> X_train, X_test, y_train, y_test = train_test_split(features, target, test_size=0.3)
-    >>> 
-    >>> model = FM(RandomForestClassifier())
-    >>> ensemble = Ensemble(quantifier=model,
-    ...                     size=50,
-    ...                     selection_metric='ptr',
-    ...                     return_type='median',
-    ...                     n_jobs=-1,
-    ...                     verbose=False)
-    >>> 
-    >>> ensemble.fit(X_train, y_train)
-    >>> 
-    >>> predictions = ensemble.predict(X_test)
-    >>> predictions
-    {0: 0.4589857954621449, 1: 0.5410142045378551}
-    >>> get_real_prev(y_test)
-    {0: 0.45614035087719296, 1: 0.543859649122807}
-    """
-        
-    SELECTION_METRICS = {'all', 'ptr', 'ds'}
-
-    def __init__(self,
-                 quantifier:Quantifier,
-                 size:int=50,
-                 min_prop:float=0.1,
-                 selection_metric:str='all',
-                 p_metric:float=0.25,
-                 return_type:str="mean",
-                 max_sample_size:int=None,
-                 max_trials:int=100, 
-                 n_jobs:int=1,
-                 verbose:bool=False):
-        
-        assert selection_metric in Ensemble.SELECTION_METRICS, \
-            f'unknown selection_metric={selection_metric}; valid are {Ensemble.SELECTION_METRICS}'
-        assert max_sample_size is None or max_sample_size > 0, \
-            'wrong value for max_sample_size; set it to a positive number or None'
-            
-        self.base_quantifier = quantifier
-        self.size = size
-        self.min_prop = min_prop
-        self.p_metric = p_metric
-        self.selection_metric = selection_metric
-        self.return_type = return_type
-        self.n_jobs = n_jobs
-        self.proba_generator = None
-        self.verbose = verbose
-        self.max_sample_size = max_sample_size
-        self.max_trials = max_trials
-
-    def sout(self, msg):
-        """Prints a message if verbose is True."""
-        if self.verbose:
-            print('[Ensemble]' + msg)
-
-    def fit(self, X, y):
-        """ Fits the ensemble model to the given training data.
-        
-        Parameters
-        ----------
-        X : array-like of shape (n_samples, n_features)
-            The input data.
-        y : array-like of shape (n_samples,)
-            The target values.
-            
-        Returns
-        -------
-        self : Ensemble
-            The fitted ensemble model.
-        """
-        self.sout('Fit')
-        
-        self.classes = np.unique(y)
-        
-        if self.selection_metric == 'ds' and not self.binary_data:
-            raise ValueError(f'ds selection_metric is only defined for binary quantification, but this dataset is not binary')
-        # randomly chooses the prevalences for each member of the ensemble (preventing classes with less than
-        # min_pos positive examples)
-        sample_size = len(y) if self.max_sample_size is None else min(self.max_sample_size, len(y))
-        prevs = [_draw_simplex(ndim=self.n_class, min_val=self.min_prop, max_trials=self.max_trials) for _ in range(self.size)]
-
-
-        posteriors = None
-        if self.selection_metric == 'ds':
-            # precompute the training posterior probabilities
-            posteriors, self.proba_generator = self.ds_get_posteriors(X, y)
-
-
-        args = (
-            (X, y, self.base_quantifier, prev, posteriors, self.verbose, sample_size)
-            for prev in prevs
-        )
-        
-        self.ensemble = parallel(
-            _delayed_new_sample,
-            tqdm(args, desc='fitting ensemble', total=self.size) if self.verbose else args,
-            n_jobs=self.n_jobs)
-
-        self.sout('Fit [Done]')
-        return self
-
-    def predict(self, X):
-        """ Predicts the class prevalences for the given test data.
-        
-        Parameters
-        ----------
-        X : array-like of shape (n_samples, n_features)
-            The input data.
-        
-        Returns
-        -------
-        prevalences : array-like of shape (n_samples, n_classes)
-            The predicted class prevalences.
-        """
-        self.sout('Predict')
-        
-        args = ((Qi, X) for Qi in self.ensemble)
-        
-        prevalences = np.asarray(
-            parallel(_delayed_predict, 
-                     tqdm(args, desc="Predicting Ensemble", total=len(self.ensemble)) if self.verbose else args, 
-                     n_jobs=self.n_jobs)
-        )
-
-        prevalences = pd.DataFrame(prevalences).to_numpy()
-        
-        self.p_metric = int(len(prevalences) * self.p_metric)
-
-        if self.selection_metric == 'ptr':
-            prevalences = self.ptr_selection_metric(prevalences)
-        elif self.selection_metric == 'ds':
-            prevalences = self.ds_selection_metric(prevalences, X)
-            
-        
-        if self.return_type == "median":   
-            prevalences = np.median(prevalences, axis=0)
-        else:      
-            prevalences = np.mean(prevalences, axis=0)
-            
-        
-        self.sout('Predict [Done]')
-        return normalize_prevalence(prevalences, self.classes)
-
-
-    def ptr_selection_metric(self, prevalences):
-        """
-        Selects the prevalence estimates from models trained on samples whose prevalence is most similar
-        to an initial approximation of the test prevalence as estimated by all models in the ensemble.
-
-        Parameters
-        ----------
-        prevalences : numpy.ndarray
-            An array of prevalence estimates provided by each model in the ensemble.
-
-        Returns
-        -------
-        numpy.ndarray
-            The selected prevalence estimates after applying the PTR selection metric.
-        """
-        test_prev_estim = prevalences.mean(axis=0)
-        tr_prevs = [m[1] for m in self.ensemble]
-        ptr_differences = [measures.mean_squared_error(test_prev_estim, ptr_i) for ptr_i in tr_prevs]
-        order = np.argsort(ptr_differences)
-        return _select_k(prevalences, order, k=self.p_metric)
-
-    def ds_get_posteriors(self, X, y):
-        """ 
-        Generate posterior probabilities using cross-validated logistic regression.
-        This method computes posterior probabilities for the training data via cross-validation,
-        using a logistic regression classifier with hyperparameters optimized through grid search.
-        It also returns a function to generate posterior probabilities for new data.
-        
-        Parameters
-        ----------
-        X : array-like of shape (n_samples, n_features)
-            The feature matrix representing the training data.
-        y : array-like of shape (n_samples,)
-            The target vector representing class labels for the training data.
-            
-        Returns
-        -------
-        posteriors : ndarray of shape (n_samples, n_classes)
-            Posterior probabilities for the training data obtained through cross-validation.
-        posteriors_generator : callable
-            A function that computes posterior probabilities for new input data.
-            
-        Notes
-        -----
-        - In scenarios where the quantifier is not based on a probabilistic classifier, it's necessary
-            to train a separate probabilistic model to obtain posterior probabilities.
-        - Using cross-validation ensures that the posterior probabilities for the training data are unbiased,
-            as each data point is evaluated by a model not trained on that point.
-        - Hyperparameters for the logistic regression classifier are optimized using a grid search with
-            cross-validation to improve the model's performance.
-        """
-        lr_base = LogisticRegression(class_weight='balanced', max_iter=1000)
-
-        optim = GridSearchCV(
-            lr_base, param_grid={'C': np.logspace(-4, 4, 9)}, cv=5, n_jobs=self.n_jobs, refit=True
-        ).fit(X, y)
-
-        posteriors = cross_val_predict(
-            optim.best_estimator_, X, y, cv=5, n_jobs=self.n_jobs, method='predict_proba'
-        )
-        posteriors_generator = optim.best_estimator_.predict_proba
-
-        return posteriors, posteriors_generator
-
-
-    def ds_selection_metric(self, prevalences, test):
-        """
-        Selects the prevalence estimates from models trained on samples whose distribution of posterior
-        probabilities is most similar to the distribution of posterior probabilities for the test data.
-        
-        Parameters
-        ----------
-        prevalences : numpy.ndarray
-            An array of prevalence estimates provided by each model in the ensemble.
-        test : array-like of shape (n_samples, n_features)
-            The feature matrix representing the test data.
-        
-        Returns
-        -------
-        numpy.ndarray
-            The selected prevalence estimates after applying the DS selection metric.
-        """
-        test_posteriors = self.proba_generator(test)
-        test_distribution = getHist(test_posteriors, 8)
-        tr_distributions = [m[2] for m in self.ensemble]
-        dist = [hellinger(tr_dist_i, test_distribution) for tr_dist_i in tr_distributions]
-        order = np.argsort(dist)
-        return _select_k(prevalences, order, k=self.p_metric)
-
-def _select_k(elements, order, k):
-    """
-    Selects the k elements from the list of elements based on the order.
-    If the list is empty, it returns the original list.
-    
-    Parameters
-    ----------
-    elements : array-like
-        The array of elements to be selected from.
-    order : array-like
-        The order of the elements.
-    k : int
-        The number of elements to be selected.
-    
-    Returns
-    -------
-    array-like
-        The selected elements.
-    """
-    elements_k = [elements[idx] for idx in order[:k]]
-    if elements_k:
-        return elements_k
-    print(f"Unable to take {k} for elements with size {len(elements)}")
-    return elements
-    
-
-
-def _delayed_new_sample(args):
-    """
-    Fits a new sample for the ensemble, this method is used for parallelization purposes generating a new artificial sample for each quantifier.
-    
-    Parameters
-    ----------
-    args : tuple
-        A tuple containing the following elements:
-        
-        X : array-like of shape (n_samples, n_features)
-            The feature matrix representing the training data.
-        y : array-like of shape (n_samples,)
-            The target vector representing class labels for the training data.
-        base_quantifier : Quantifier
-            The base quantifier model to be used in the ensemble.
-        prev : array-like of shape (n_classes,)
-            The class prevalences for the new sample.
-        posteriors : array-like of shape (n_samples, n_classes)
-            The posterior probabilities for the training data obtained through cross-validation.
-        verbose : bool
-            If True, prints progress messages during fitting and prediction.
-        sample_size : int
-            The size of the sample to be generated.
-    
-    Returns
-    -------
-    tuple
-        A tuple containing the following elements:
-        
-        model : Quantifier
-            The fitted quantifier model.
-        tr_prevalence : array-like of shape (n_classes,)
-            The class prevalences for the new sample.
-        tr_distribution : array-like of shape (n_classes,)
-            The distribution of posterior probabilities for the new sample.
-        X : array-like of shape (n_samples, n_features)
-            The feature matrix representing the training data.
-        y : array-like of shape (n_samples,)
-            The target vector representing class labels for the training data.
-    """
-    
-    
-    X, y, base_quantifier, prev, posteriors, verbose, sample_size = args
-    if verbose:
-        print(f'\tfit-start for prev {str(np.round(prev, 3))}, sample_size={sample_size}')
-    model = deepcopy(base_quantifier)
-
-    sample_index = get_indexes_with_prevalence(y, prev, sample_size)
-    X_sample = np.take(X, sample_index, axis=0)
-    y_sample = np.take(y, sample_index, axis=0)
-    #print(X_sample)
-
-    model.fit(X_sample, y_sample)
-
-    tr_prevalence = prev
-    tr_distribution = getHist(posteriors[sample_index], 8) if (posteriors is not None) else None
-    if verbose:
-        print(f'\t \\--fit-ended for prev {str(np.round(prev, 3))}')
-    return (model, tr_prevalence, tr_distribution, X, y)
-
-
-def _delayed_predict(args):
-    """
-    Predicts the class prevalences for the given test data.
-    
-    Parameters
-    ----------
-    args : tuple
-        A tuple containing the following elements:
-        
-        quantifier : Quantifier
-            The quantifier model to be used for prediction.
-        X : array-like of shape (n_samples, n_features)
-            The input data.
-    
-    Returns
-    -------
-    array-like of shape (n_samples, n_classes)
-        The predicted class prevalences.
-    """
-    quantifier, X = args
-    #print(np.asarray(list(quantifier[0].predict(X).values())))
-    return list(quantifier[0].predict(X).values())
-
-
-def _draw_simplex(ndim, min_val, max_trials=100):
-    """
-    Return a uniform sample from the ndim-dimensional simplex, ensuring all dimensions are >= min_val.
-
-    Note:
-        For min_val > 0, the sampling is not truly uniform because the simplex is restricted.
-
-    Parameters:
-        ndim (int): Number of dimensions of the simplex.
-        min_val (float): Minimum allowed value for each dimension. Must be less than 1 / ndim.
-        max_trials (int, optional): Maximum number of attempts to find a valid sample (default is 100).
-
-    Returns:
-        numpy.ndarray: A sample from the ndim-dimensional simplex where all dimensions are >= min_val.
-
-    Raises:
-        ValueError: If min_val >= 1 / ndim, or if a valid sample cannot be found within max_trials trials.
-    """
-    if min_val >= 1 / ndim:
-        raise ValueError(f'no sample can be draw from the {ndim}-dimensional simplex so that '
-                         f'all its values are >={min_val} (try with a larger value for min_pos)')
-    trials = 0
-    while True:
-        u = make_prevs(ndim)
-        if all(u >= min_val):
-            return u
-        trials += 1
-        if trials >= max_trials:
-            raise ValueError(f'it looks like finding a random simplex with all its dimensions being'
-                             f'>= {min_val} is unlikely (it failed after {max_trials} trials)')
-