|
|
from abc import ABC, abstractmethod |
|
|
from typing import Tuple, List |
|
|
|
|
|
import numpy as np |
|
|
|
|
|
from src.dto.dto import ExplanationDto |
|
|
|
|
|
|
|
|
class Segregator(ABC): |
|
|
@abstractmethod |
|
|
def segregate( |
|
|
self, explanations: ExplanationDto |
|
|
) -> Tuple[List[str], List[str], List[str]]: |
|
|
... |
|
|
|
|
|
|
|
|
class PercentileBasedSegregator(Segregator): |
|
|
def __init__( |
|
|
self, |
|
|
upper_bound_percentile: int = 85, |
|
|
middle_bound_percentile: int = 75, |
|
|
lower_bound_percentile: int = 10, |
|
|
): |
|
|
self.__upper_bound_percentile = upper_bound_percentile |
|
|
self.__middle_bound_percentile = middle_bound_percentile |
|
|
self.__lower_bound_percentile = lower_bound_percentile |
|
|
|
|
|
def segregate( |
|
|
self, |
|
|
explanations: ExplanationDto, |
|
|
) -> Tuple[List[str], List[str], List[str]]: |
|
|
scores = [explanation.score for explanation in explanations.explanations] |
|
|
scores = np.asarray(scores) |
|
|
upper_bound = np.percentile(scores, self.__upper_bound_percentile) |
|
|
mid_bound = np.percentile(scores, self.__middle_bound_percentile) |
|
|
lower_bound = np.percentile(scores, self.__lower_bound_percentile) |
|
|
|
|
|
pos_features = [ |
|
|
explanation.feature |
|
|
for explanation in explanations.explanations |
|
|
if explanation.score >= upper_bound and explanation.score != 0 |
|
|
] |
|
|
mid_features = [ |
|
|
explanation.feature |
|
|
for explanation in explanations.explanations |
|
|
if upper_bound > explanation.score >= mid_bound > 0 |
|
|
] |
|
|
low_features = [ |
|
|
explanation.feature |
|
|
for explanation in explanations.explanations |
|
|
if mid_bound > explanation.score >= lower_bound > 0 |
|
|
] |
|
|
|
|
|
return pos_features, mid_features, low_features |
|
|
|
