eckity.sklearn_compatible.regression_evaluator

 1"""
 2This module implements the fitness evaluation class, which delivers the fitness function.
 3You will need to implement such a class to work with your own problem and fitness function.
 4"""
 5
 6from sklearn.metrics import mean_absolute_error
 7
 8from eckity.evaluators.simple_individual_evaluator import SimpleIndividualEvaluator
 9
10
11class RegressionEvaluator(SimpleIndividualEvaluator):
12    """
13    Computes the fitness of an individual in regression problems.
14
15    Parameters
16    ----------
17    X: array-like of shape (n_samples, n_features), default=None
18    Training/Test data.
19
20    y: array-like of shape (n_samples,) or (n_samples, 1), default=None
21    Target vector. used during the training phase.
22
23    metric: callable (optional, default=mean_absolute_error)
24    A function which receives two array-like of shapes (n_samples,) or (n_samples, 1) and returns a float or
25    ndarray of floats
26    """
27        
28    def __init__(self, X=None, y=None, metric=mean_absolute_error):
29        super().__init__()
30        self.X = X
31        self.y = y
32        self.metric = metric
33
34    def set_context(self, context):
35        """
36        Receive X and y values and assign them to X and y fields.
37
38        Parameters
39        ----------
40        context: tuple. first element is a numpy array of size (n_samples, n_features),
41                        and the second element is a numpy array of size (n_samples, 1) or (n_samples,)
42            X matrix and y vector, either (X_train, y_train) or (X_test, None), depending on the evolution stage
43
44        Returns
45        -------
46        None.
47
48        Examples
49        -------
50        reg_eval = RegressionEvaluator()
51        X, y = make_regression()
52        X_train, X_test, y_train, y_test = train_test_split()
53        reg_eval.set_context(X_train, y_train)
54        """
55        self.X = context[0]
56        self.y = context[1]
57
58    def evaluate_individual(self, individual):
59        """
60        compute fitness value by computing the MAE between program tree execution result and y result vector
61
62        Parameters
63        ----------
64        individual : Tree
65            An individual program tree in the GP population, whose fitness needs to be computed.
66            Makes use of GPTree.execute, which runs the program.
67            In Sklearn settings, calling `individual.execute` must use a numpy array.
68            For example, if self.X is X_train/X_test, the call is `individual.execute(self.X)`.
69
70        Returns
71        ----------
72        float
73            Computed fitness value - evaluated using the provided scoring function between the execution result of X and
74            the vector y.
75        """
76        return self.metric(self.y, individual.execute(self.X))