kgcnn.metrics package¶
Submodules¶
kgcnn.metrics.metrics module¶
-
class
kgcnn.metrics.metrics.
AUCNoNaN
(name='AUC_no_nan', **kwargs)[source]¶ Bases:
keras.src.metrics.confusion_metrics.AUC
-
update_state
(y_true, y_pred, sample_weight=None)[source]¶ Accumulates confusion matrix statistics.
- Parameters
y_true – The ground truth values.
y_pred – The predicted values.
sample_weight – Optional weighting of each example. Can be a tensor whose rank is either 0, or the same rank as y_true, and must be broadcastable to y_true. Defaults to 1.
-
-
class
kgcnn.metrics.metrics.
BalancedBinaryAccuracyNoNaN
(name='balanced_binary_accuracy_no_nan', class_id=None, num_thresholds=1, specificity=0.5, **kwargs)[source]¶ Bases:
keras.src.metrics.confusion_metrics.SensitivityAtSpecificity
-
result
()[source]¶ Compute the current metric value.
- Returns
A scalar tensor, or a dictionary of scalar tensors.
-
update_state
(y_true, y_pred, sample_weight=None)[source]¶ Update the state of the metric.
- Parameters
y_true – Ground truth label values. shape = [batch_size, d0, .. dN-1] or shape = [batch_size, d0, .. dN-1, 1] .
y_pred – The predicted probability values. shape = [batch_size, d0, .. dN] .
sample_weight – Optional sample_weight acts as a coefficient for the metric.
-
-
class
kgcnn.metrics.metrics.
BinaryAccuracyNoNaN
(name='binary_accuracy_no_nan', dtype=None, threshold=0.5, **kwargs)[source]¶ Bases:
keras.src.metrics.reduction_metrics.MeanMetricWrapper
-
class
kgcnn.metrics.metrics.
ScaledForceMeanAbsoluteError
(scaling_shape=(1, 1), name='force_mean_absolute_error', dtype_scale: Optional[str] = None, squeeze_states: bool = True, find_padded_atoms: bool = True, **kwargs)[source]¶ Bases:
keras.src.metrics.reduction_metrics.MeanMetricWrapper
Metric for a scaled mean absolute error (MAE), which can undo a pre-scaling of the targets. Only intended as metric this allows to info the MAE with correct units or absolute values during fit.
-
class
kgcnn.metrics.metrics.
ScaledMeanAbsoluteError
(scaling_shape=(), name='mean_absolute_error', dtype_scale: Optional[str] = None, **kwargs)[source]¶ Bases:
keras.src.metrics.regression_metrics.MeanAbsoluteError
Metric for a scaled mean absolute error (MAE), which can undo a pre-scaling of the targets. Only intended as metric this allows to info the MAE with correct units or absolute values during fit.
-
class
kgcnn.metrics.metrics.
ScaledRootMeanSquaredError
(scaling_shape=(), name='root_mean_squared_error', dtype_scale: Optional[str] = None, **kwargs)[source]¶ Bases:
keras.src.metrics.regression_metrics.RootMeanSquaredError
Metric for a scaled root mean squared error (RMSE), which can undo a pre-scaling of the targets. Only intended as metric this allows to info the MAE with correct units or absolute values during fit.
-
reset_state
()[source]¶ Reset all of the metric state variables.
This function is called between epochs/steps, when a metric is evaluated during training.
-
update_state
(y_true, y_pred, sample_weight=None)[source]¶ Accumulates root mean squared error statistics.
- Parameters
y_true – The ground truth values.
y_pred – The predicted values.
sample_weight – Optional weighting of each example. Can be a Tensor whose rank is either 0, or the same rank as y_true, and must be broadcastable to y_true. Defaults to 1.
- Returns
Update op.
-