Module miplearn.classifiers.threshold

Expand source code 
#  MIPLearn: Extensible Framework for Learning-Enhanced Mixed-Integer Optimization
#  Copyright (C) 2020, UChicago Argonne, LLC. All rights reserved.
#  Released under the modified BSD license. See COPYING.md for more details.

from abc import abstractmethod, ABC

import numpy as np
from sklearn.metrics._ranking import _binary_clf_curve

from miplearn.classifiers import Classifier


class DynamicThreshold(ABC):
    @abstractmethod
    def find(
        self,
        clf: Classifier,
        x_train: np.ndarray,
        y_train: np.ndarray,
    ) -> float:
        """
        Given a trained binary classifier `clf` and a training data set,
        returns the numerical threshold (float) satisfying some criterea.
        """
        pass


class MinPrecisionThreshold(DynamicThreshold):
    """
    The smallest possible threshold satisfying a minimum acceptable true
    positive rate (also known as precision).
    """

    def __init__(self, min_precision: float) -> None:
        self.min_precision = min_precision

    def find(self, clf, x_train, y_train):
        proba = clf.predict_proba(x_train)

        assert isinstance(proba, np.ndarray), "classifier should return numpy array"
        assert proba.shape == (
            x_train.shape[0],
            2,
        ), "classifier should return (%d,%d)-shaped array, not %s" % (
            x_train.shape[0],
            2,
            str(proba.shape),
        )

        fps, tps, thresholds = _binary_clf_curve(y_train, proba[:, 1])
        precision = tps / (tps + fps)

        for k in reversed(range(len(precision))):
            if precision[k] >= self.min_precision:
                return thresholds[k]
        return 2.0

Classes

class DynamicThreshold (*args, **kwargs)

Helper class that provides a standard way to create an ABC using inheritance.

Expand source code 
class DynamicThreshold(ABC):
    @abstractmethod
    def find(
        self,
        clf: Classifier,
        x_train: np.ndarray,
        y_train: np.ndarray,
    ) -> float:
        """
        Given a trained binary classifier `clf` and a training data set,
        returns the numerical threshold (float) satisfying some criterea.
        """
        pass

Ancestors

  • abc.ABC

Subclasses

Methods

def find(self, clf, x_train, y_train)

Given a trained binary classifier clf and a training data set, returns the numerical threshold (float) satisfying some criterea.

Expand source code 
@abstractmethod
def find(
    self,
    clf: Classifier,
    x_train: np.ndarray,
    y_train: np.ndarray,
) -> float:
    """
    Given a trained binary classifier `clf` and a training data set,
    returns the numerical threshold (float) satisfying some criterea.
    """
    pass
class MinPrecisionThreshold (min_precision)

The smallest possible threshold satisfying a minimum acceptable true positive rate (also known as precision).

Expand source code 
class MinPrecisionThreshold(DynamicThreshold):
    """
    The smallest possible threshold satisfying a minimum acceptable true
    positive rate (also known as precision).
    """

    def __init__(self, min_precision: float) -> None:
        self.min_precision = min_precision

    def find(self, clf, x_train, y_train):
        proba = clf.predict_proba(x_train)

        assert isinstance(proba, np.ndarray), "classifier should return numpy array"
        assert proba.shape == (
            x_train.shape[0],
            2,
        ), "classifier should return (%d,%d)-shaped array, not %s" % (
            x_train.shape[0],
            2,
            str(proba.shape),
        )

        fps, tps, thresholds = _binary_clf_curve(y_train, proba[:, 1])
        precision = tps / (tps + fps)

        for k in reversed(range(len(precision))):
            if precision[k] >= self.min_precision:
                return thresholds[k]
        return 2.0

Ancestors

Inherited members