Merge pull request #2263 from hsjang001205/c_index (d79c5ea4) · Commits · 钟慕尧 / deepchem

deepchem/metrics/init.py

+1 −0

Original line number	Diff line number	Diff line
		@@ -37,3 +37,4 @@ from deepchem.metrics.score_function import prc_auc_score
		from deepchem.metrics.score_function import rms_score
		from deepchem.metrics.score_function import mae_score
		from deepchem.metrics.score_function import bedroc_score
		from deepchem.metrics.score_function import concordance_index

+7 −15

Original line number	Diff line number	Diff line
		@@ -528,20 +528,11 @@ class Metric(object):
		if mode is None:
		# These are some smart defaults
		if self.metric.__name__ in [
		"roc_auc_score",
		"matthews_corrcoef",
		"recall_score",
		"accuracy_score",
		"kappa_score",
		"cohen_kappa_score",
		"precision_score",
		"balanced_accuracy_score",
		"prc_auc_score",
		"f1_score",
		"bedroc_score",
		"jaccard_score",
		"jaccard_index",
		"pixel_error",
		"roc_auc_score", "matthews_corrcoef", "recall_score",
		"accuracy_score", "kappa_score", "cohen_kappa_score",
		"precision_score", "balanced_accuracy_score", "prc_auc_score",
		"f1_score", "bedroc_score", "jaccard_score", "jaccard_index",
		"pixel_error"
		]:
		mode = "classification"
		# These are some smart defaults corresponding to sklearn's required
		@@ -561,7 +552,8 @@ class Metric(object):
		classification_handling_mode = None
		elif self.metric.__name__ in [
		"pearson_r2_score", "r2_score", "mean_squared_error",
		"mean_absolute_error", "rms_score", "mae_score", "pearsonr"
		"mean_absolute_error", "rms_score", "mae_score", "pearsonr",
		"concordance_index"
		]:
		mode = "regression"
		else:

+52 −0

Original line number	Diff line number	Diff line
		@@ -162,3 +162,55 @@ def bedroc_score(y_true: np.ndarray, y_pred: np.ndarray, alpha: float = 20.0):
		scores = sorted(scores, key=lambda pair: pair[1], reverse=True)

		return CalcBEDROC(scores, 0, alpha)


		def concordance_index(y_true: np.ndarray, y_pred: np.ndarray) -> float:
		"""Compute Concordance index.

		Statistical metric indicates the quality of the predicted ranking.
		Please confirm details from [1]_.

		Parameters
		----------
		y_true: np.ndarray
		continous value
		y_pred: np.ndarray
		Predicted value

		Returns
		-------
		float
		score between [0,1]

		References
		----------
		.. [1] Steck, Harald, et al. "On ranking in survival analysis:
		Bounds on the concordance index." Advances in neural information processing systems (2008): 1209-1216.
		"""

		idx = np.argsort(y_true)
		y_true = y_true[idx]
		y_pred = y_pred[idx]

		pairs = 0
		correct_pairs = 0.0

		for i in range(len(y_true)):
		true_a = y_true[i]
		pred_a = y_pred[i]

		for j in range(i + 1, len(y_true)):
		true_b = y_true[j]
		pred_b = y_pred[j]
		if true_a != true_b:
		pairs += 1
		if pred_a == pred_b:
		correct_pairs += 0.5
		elif pred_a < pred_b:
		correct_pairs += true_a < true_b
		else:
		correct_pairs += true_a > true_b

		assert pairs > 0, 'No pairs for comparision'

		return correct_pairs / pairs

+20 −0

Original line number	Diff line number	Diff line
		@@ -68,3 +68,23 @@ def test_bedroc_score():
		np.concatenate([worst_pred_actives, worst_pred_inactives]))
		worst_score = dc.metrics.bedroc_score(y_true, y_pred_worst)
		np.testing.assert_almost_equal(worst_score, 0.0, 4)


		def test_concordance_index():
		"""Test concordance index."""

		metric = dc.metrics.Metric(dc.metrics.concordance_index)

		y_true = np.array([1, 3, 5, 4, 2])
		y_pred = np.array([3, 1, 5, 4, 2])
		assert metric.compute_singletask_metric(y_true, y_pred) == 0.7

		# best case
		y_true = np.array([1, 3, 5, 4, 2])
		y_pred = np.array([1, 3, 5, 4, 2])
		assert metric.compute_singletask_metric(y_true, y_pred) == 1.0

		# duplicate prediction value
		y_true = np.array([1, 3, 5, 4, 2])
		y_pred = np.array([1, 3, 4, 4, 2])
		assert metric.compute_singletask_metric(y_true, y_pred) == 0.95

+2 −0

Original line number	Diff line number	Diff line
		@@ -79,6 +79,8 @@ DeepChem has a variety of different metrics which are useful for measuring model

		.. autofunction:: deepchem.metrics.bedroc_score

		.. autofunction:: deepchem.metrics.concordance_index

		.. autofunction:: deepchem.metrics.genomic_metrics.get_motif_scores

		.. autofunction:: deepchem.metrics.genomic_metrics.get_pssm_scores