import numpy as np
from caliber.multiclass_classification.base import AbstractMulticlassClassificationModel
from caliber.multiclass_classification.pred_from_probs_mixin import (
PredFromProbsMulticlassClassificationMixin,
)
[docs]
class HistogramBinningMulticlassClassificationModel(
PredFromProbsMulticlassClassificationMixin, AbstractMulticlassClassificationModel
):
def __init__(self, n_prob_bins: int = 10, min_prob_bin: float = 0.0):
super().__init__()
self.n_prob_bins = n_prob_bins
self._min_prob_bin = min_prob_bin
self._prob_bin_edges = None
[docs]
def fit(self, probs: np.ndarray, targets: np.ndarray):
self._n_classes = probs.shape[1]
self._prob_bin_edges = self._get_prob_bin_edges()
prob_bin_indices = np.digitize(probs, self._prob_bin_edges)
top_class_indices = np.argmax(probs, axis=1)
self._params = np.empty((self.n_prob_bins + 1, self._n_classes))
for i in range(1, self.n_prob_bins + 2):
for c in range(self._n_classes):
mask = self._get_mask(i, c, prob_bin_indices, top_class_indices)
self._fit_bin(i, c, mask, targets)
[docs]
def predict_proba(self, probs: np.ndarray) -> np.ndarray:
if self._prob_bin_edges is None:
raise ValueError("Run `fit` first.")
if probs.shape[1] != self._n_classes:
raise ValueError(
"The number of classes when fitting and predicting must be the same."
)
prob_bin_indices = np.digitize(probs, self._prob_bin_edges)
top_class_indices = np.argmax(probs, axis=1)
probs = np.copy(probs)
for i in range(1, self.n_prob_bins + 2):
for c in range(0, self._n_classes):
mask = self._get_mask(i, c, prob_bin_indices, top_class_indices)
if not np.isnan(self._params[i - 1, c]):
probs[mask, c] = self._params[i - 1, c]
return probs
def _get_prob_bin_edges(self) -> np.ndarray:
return np.linspace(0, 1, self.n_prob_bins + 1)
def _fit_bin(self, i: int, c: int, mask: np.ndarray, targets: np.ndarray):
prob_bin = np.mean(mask)
if prob_bin >= self._min_prob_bin:
self._params[i - 1, c] = np.mean(targets[mask] == c)
else:
self._params[i - 1, c] = np.nan
@staticmethod
def _get_mask(
prob_bin_idx: int,
class_idx: int,
prob_bin_indices: np.ndarray,
top_class_indices: np.ndarray,
) -> np.ndarray:
return (prob_bin_indices[:, class_idx] == prob_bin_idx) * (
top_class_indices == class_idx
)