mutated_validation.py

"""implementation of Mutation  Validation Score"""

from enum import Enum
from typing import Union

from dataclasses import dataclass
import random
from sklearn.metrics import (
    accuracy_score,
    precision_score,
    roc_auc_score,
    recall_score
)
from imblearn.metrics import geometric_mean_score
import numpy as np

from mutate_labels import MutatedValidation

random.seed(42)
np.random.seed(42)


class EvaluationMetricsType(str, Enum):
    ACCURACY = "accuracy"
    PRECISION = "precision"
    RECALL = "recall"
    ROC_AUC = "roc_auc"
    G_MEAN = "g_mean"


class TrainRun(int, Enum):
    ORIGINAL = 0
    MUTATED = 1


@dataclass
class EvaluationMetric:
    predicted_labels: np.ndarray
    original_labels: np.ndarray
    metric_score: Union[float, int]


def get_metric_score(
        predicted_labels: np.ndarray, original_labels: np.ndarray, metric: str
) -> EvaluationMetric:
    score = ""

    if metric == EvaluationMetricsType.ACCURACY:
        score = accuracy_score(original_labels, predicted_labels)
    if metric == EvaluationMetricsType.RECALL:
        score = recall_score(original_labels, predicted_labels)
    if metric == EvaluationMetricsType.PRECISION:
        score = precision_score(original_labels, predicted_labels)
    if metric == EvaluationMetricsType.ROC_AUC:
        score = roc_auc_score(original_labels, predicted_labels)
    if metric == EvaluationMetricsType.G_MEAN:
        score = geometric_mean_score(original_labels, predicted_labels)
    return EvaluationMetric(
        predicted_labels=predicted_labels,
        original_labels=original_labels,
        metric_score=score,  # type: ignore
    )


@dataclass
class MutatedValidationScore:
    mutated_labels: MutatedValidation
    mutate: np.ndarray
    original_training_predicted_labels: np.ndarray
    mutated_training_predicted_labels: np.ndarray
    evaluation_metric: str

    @property
    def mutated_training_metric_score_based_original_labels(self):
        return get_metric_score(
            predicted_labels=self.mutated_training_predicted_labels,
            original_labels=self.mutated_labels.labels,
            metric=self.evaluation_metric,
        )

    @property
    def original_training_metric_score_based_on_original_labels(self):
        return get_metric_score(
            predicted_labels=self.original_training_predicted_labels,
            original_labels=self.mutated_labels.labels,
            metric=self.evaluation_metric,
        )

    @property
    def mutated_training_metric_score_based_mutated_labels(self):
        return get_metric_score(
            predicted_labels=self.mutated_training_predicted_labels,
            original_labels=self.mutate,
            metric=self.evaluation_metric,
        )

    @property
    def get_mv_score(self):
        return (
                (
                        (1 - 2 * self.mutated_labels.perturbation_ratio)
                        * self.mutated_training_metric_score_based_original_labels.metric_score
                )
                + (
                        self.original_training_metric_score_based_on_original_labels.metric_score
                        - self.mutated_training_metric_score_based_mutated_labels.metric_score
                )
                + self.mutated_labels.perturbation_ratio
        )