Loading deepchem/metrics/__init__.py +12 −1 Original line number Diff line number Diff line Loading @@ -55,6 +55,16 @@ def compute_roc_auc_scores(y, y_pred): return score def balanced_accuracy_score(y, y_pred): """Computes balanced accuracy score.""" num_positive = float(np.count_nonzero(y)) num_negative = float(len(y) - num_positive) pos_weight = num_positive/num_negative weights = np.ones_like(y) weights[y != 0] = pos_weight return accuracy_score(y, y_pred, sample_weight=weights) def pearson_r2_score(y, y_pred): """Computes Pearson R^2 (square of Pearson correlation).""" return pearsonr(y, y_pred)[0]**2 Loading Loading @@ -137,7 +147,8 @@ class Metric(object): if mode is None: if self.metric.__name__ in [ "roc_auc_score", "matthews_corrcoef", "recall_score", "accuracy_score", "kappa_score", "precision_score" "accuracy_score", "kappa_score", "precision_score", "balanced_accuracy_score" ]: mode = "classification" elif self.metric.__name__ in [ Loading Loading
deepchem/metrics/__init__.py +12 −1 Original line number Diff line number Diff line Loading @@ -55,6 +55,16 @@ def compute_roc_auc_scores(y, y_pred): return score def balanced_accuracy_score(y, y_pred): """Computes balanced accuracy score.""" num_positive = float(np.count_nonzero(y)) num_negative = float(len(y) - num_positive) pos_weight = num_positive/num_negative weights = np.ones_like(y) weights[y != 0] = pos_weight return accuracy_score(y, y_pred, sample_weight=weights) def pearson_r2_score(y, y_pred): """Computes Pearson R^2 (square of Pearson correlation).""" return pearsonr(y, y_pred)[0]**2 Loading Loading @@ -137,7 +147,8 @@ class Metric(object): if mode is None: if self.metric.__name__ in [ "roc_auc_score", "matthews_corrcoef", "recall_score", "accuracy_score", "kappa_score", "precision_score" "accuracy_score", "kappa_score", "precision_score", "balanced_accuracy_score" ]: mode = "classification" elif self.metric.__name__ in [ Loading
