Changes (6b467aa2) · Commits · 钟慕尧 / deepchem

deepchem/metrics/init.py

+7 −77

Original line number	Diff line number	Diff line
		@@ -4,7 +4,6 @@ import numpy as np
		import warnings
		import sklearn.metrics
		import logging
		# TODO: Imported metrics will be removed in a futrue version of DeepCHem
		from sklearn.metrics import matthews_corrcoef
		from sklearn.metrics import recall_score
		from sklearn.metrics import r2_score
		@@ -20,76 +19,6 @@ from scipy.stats import pearsonr
		logger = logging.getLogger(__name__)


		def matthews_corrcoef(args, *kwargs):
		logger.warning(
		"matthews_corrcoef is deprecated. Use sklearn.metrics.matthews_corrcoef instead. dc.metrics.matthews_corrcoef will be removed in a future version of DeepChem."
		)
		return sklearn.metrics.matthews_corrcoef(args, *kwargs)


		def recall_score(args, *kwargs):
		logger.warning(
		"recall_score is deprecated. Use sklearn.metrics.recall_score instead. dc.metrics.recall_score will be removed in a future version of DeepChem."
		)
		return sklearn.metrics.recall_score(args, *kwargs)


		def r2_score(args, *kwargs):
		logger.warning(
		"r2_score is deprecated. Use sklearn.metrics.r2_score instead. dc.metrics.r2_score will be removed in a future version of DeepChem."
		)
		return sklearn.metrics.r2_score(args, *kwargs)


		def mean_squared_error(args, *kwargs):
		logger.warning(
		"mean_squared_error is deprecated. Use sklearn.metrics.mean_squared_error instead. dc.metrics.mean_squared_error will be removed in a future version of DeepChem."
		)
		return sklearn.metrics.mean_squared_error(args, *kwargs)


		def mean_absolute_error(args, *kwargs):
		logger.warning(
		"mean_absolute_error is deprecated. Use sklearn.metrics.mean_absolute_error instead. dc.metrics.mean_absolute_error will be removed in a future version of DeepChem."
		)
		return sklearn.metrics.mean_absolute_error(args, *kwargs)


		def precision_score(args, *kwargs):
		logger.warning(
		"precision_score is deprecated. Use sklearn.metrics.precision_score instead. dc.metrics.precision_score will be removed in a future version of DeepChem."
		)
		return sklearn.metrics.precision_score(args, *kwargs)


		def precision_recall_curve(args, *kwargs):
		logger.warning(
		"precision_recall_curve is deprecated. Use sklearn.metrics.precision_recall_curve instead. dc.metrics.precision_recall_curve will be removed in a future version of DeepChem."
		)
		return sklearn.metrics.precision_recall_curve(args, *kwargs)


		def auc(args, *kwargs):
		logger.warning(
		"auc is deprecated. Use sklearn.metrics.auc instead. dc.metrics.auc will be removed in a future version of DeepChem."
		)
		return sklearn.metrics.auc(args, *kwargs)


		def jaccard_score(args, *kwargs):
		logger.warning(
		"jaccard_score is deprecated. Use sklearn.metrics.jaccard_score instead. dc.metrics.jaccard_score will be removed in a future version of DeepChem."
		)
		return sklearn.metrics.jaccard_score(args, *kwargs)


		def f1_score(args, *kwargs):
		logger.warning(
		"f1_score is deprecated. Use sklearn.metrics.f1_score instead. dc.metrics.f1_score will be removed in a future version of DeepChem."
		)
		return sklearn.metrics.f1_score(args, *kwargs)


		def threshold_predictions(y, threshold=0.5):
		"""Threshold predictions from classification model.

		@@ -135,7 +64,7 @@ def normalize_weight_shape(w, n_samples, n_tasks):
		w: np.ndarray
		`w` can be `None` or a scalar or a `np.ndarray` of shape
		`(n_samples,)` or of shape `(n_samples, n_tasks)`. If `w` is a
		sclar, it's assumed to be the same weight for all samples/tasks.
		scalar, it's assumed to be the same weight for all samples/tasks.
		n_samples: int
		The number of samples in the dataset. If `w` is not None, we should
		have `n_samples = w.shape[0]` if `w` is a ndarray
		@@ -198,7 +127,7 @@ def normalize_prediction_shape(y, mode=None, n_classes=None):
		----------
		y: np.ndarray
		If `mode=="classification"`, `y` is an array of shape `(N,)` or
		`(N, n_classes)` or `(N, n_tasks, n_classes)`. If `y` is of shape
		`(N, n_classes)` or `(N, n_tasks, n_classes)`. If `y` is an array of shape
		`(N,)` in order to impute the number of classes correctly, `y`
		must take values from `0` to `n_classes-1` as integers. If
		`mode=="regression"`, `y` is an array of shape `(N,)` or `(N,
		@@ -265,12 +194,12 @@ def normalize_prediction_shape(y, mode=None, n_classes=None):
		elif len(y.shape) == 3:
		if y.shape[-1] != 1:
		raise ValueError(
		"y must a float sclar or a ndarray of shape `(N,)` or `(N, n_tasks)` or `(N, n_tasks, 1)` for regression problems."
		"y must a float scalar or a ndarray of shape `(N,)` or `(N, n_tasks)` or `(N, n_tasks, 1)` for regression problems."
		)
		y_out = np.squeeze(y, axis=-1)
		else:
		raise ValueError(
		"y must a float sclar or a ndarray of shape `(N,)` or `(N, n_tasks)` or `(N, n_tasks, 1)` for regression problems."
		"y must a float scalar or a ndarray of shape `(N,)` or `(N, n_tasks)` or `(N, n_tasks, 1)` for regression problems."
		)
		else:
		# In this clase, y is a scalar.
		@@ -278,14 +207,14 @@ def normalize_prediction_shape(y, mode=None, n_classes=None):
		y = float(y)
		except TypeError:
		raise ValueError(
		"y must a float sclar or a ndarray of shape `(N,)` or `(N, n_tasks)` or `(N, n_tasks, 1)` for regression problems."
		"y must a float scalar or a ndarray of shape `(N,)` or `(N, n_tasks)` or `(N, n_tasks, 1)` for regression problems."
		)
		y = np.array(y)
		y_out = np.reshape(y, (1, 1))
		else:
		# If mode isn't classification or regression don't perform any
		# transformations.
		y_out = y
		raise ValueError("mode must be either classification or regression.")
		return y_out


		@@ -688,6 +617,7 @@ class Metric(object):
		if n_tasks == 1:
		computed_metrics = computed_metrics[0]

		# DEPRECATED. WILL BE REMOVED IN NEXT DEEPCHEM VERSION
		if filter_nans:
		computed_metrics = np.array(computed_metrics)
		computed_metrics = computed_metrics[~np.isnan(computed_metrics)]

deepchem/models/sklearn_models/init.py

+3 −1

Original line number	Diff line number	Diff line
		@@ -33,7 +33,9 @@ class SklearnModel(Model):
		reason for this might be that you want to do an apples-to-apples
		comparison of a scikit-learn model to another DeepChem model, or
		perhaps you want to use the hyperparameter tuning capabilities in
		`dc.hyper`. The `SklearnModel` class provides a
		`dc.hyper`. The `SklearnModel` class provides a wrapper around scikit-learn
		models that allows scikit-learn models to be trained on `Dataset` objects
		and evaluated with the same metrics as other DeepChem models.`
		"""

		def __init__(self, model_instance=None, model_dir=None, **kwargs):

deepchem/utils/evaluate.py

+6 −6

Original line number	Diff line number	Diff line
		@@ -51,14 +51,14 @@ def output_predictions(dataset, y_preds, csv_out):
		csv_out: str
		Name of file to write predictions to.
		"""
		mol_ids = dataset.ids
		data_ids = dataset.ids
		n_tasks = len(dataset.get_task_names())
		y_preds = np.reshape(y_preds, (len(y_preds), n_tasks))
		assert len(y_preds) == len(mol_ids)
		assert len(y_preds) == len(data_ids)
		with open(csv_out, "w") as csvfile:
		csvwriter = csv.writer(csvfile)
		csvwriter.writerow(["ID"] + dataset.get_task_names())
		for mol_id, y_pred in zip(mol_ids, y_preds):
		for mol_id, y_pred in zip(data_ids, y_preds):
		csvwriter.writerow([mol_id] + list(y_pred))


		@@ -218,14 +218,14 @@ class Evaluator(object):
		logger.warning(
		"Evaluator.output_predictions is deprecated. Please use dc.utils.evaluate.output_predictions instead. This method will be removed in a future version of DeepChem."
		)
		mol_ids = self.dataset.ids
		data_ids = self.dataset.ids
		n_tasks = len(self.dataset.get_task_names())
		y_preds = np.reshape(y_preds, (len(y_preds), n_tasks))
		assert len(y_preds) == len(mol_ids)
		assert len(y_preds) == len(data_ids)
		with open(csv_out, "w") as csvfile:
		csvwriter = csv.writer(csvfile)
		csvwriter.writerow(["ID"] + self.dataset.get_task_names())
		for mol_id, y_pred in zip(mol_ids, y_preds):
		for mol_id, y_pred in zip(data_ids, y_preds):
		csvwriter.writerow([mol_id] + list(y_pred))

		def compute_model_performance(self,

docs/metrics.rst

+24 −0

Original line number	Diff line number	Diff line
		@@ -9,6 +9,8 @@ depending on the type of model at hand.

		Metric Utilities
		----------------
		Metric utility functions allow for some common manipulations such as
		switching to/from one-hot representations.

		.. autofunction:: deepchem.metrics.to_one_hot

		@@ -17,6 +19,28 @@ Metric Utilities

		Metric Functions
		----------------
		DeepChem has a variety of different metrics which are useful for measuring model performance. A number (but not all) of these metrics are directly sourced from :code:`sklearn`.

		.. autofunction:: deepchem.metrics.matthews_corrcoef

		.. autofunction:: deepchem.metrics.recall_score

		.. autofunction:: deepchem.metrics.r2_score

		.. autofunction:: deepchem.metrics.mean_squared_error

		.. autofunction:: deepchem.metrics.mean_absolute_error

		.. autofunction:: deepchem.metrics.precision_score

		.. autofunction:: deepchem.metrics.precision_recall_curve

		.. autofunction:: deepchem.metrics.auc

		.. autofunction:: deepchem.metrics.jaccard_score

		.. autofunction:: deepchem.metrics.f1_score

		.. autofunction:: deepchem.metrics.roc_auc_score

		.. autofunction:: deepchem.metrics.accuracy_score

Admin message