obnb.metric

Standard metrics

Standard metric extending those available in sklearn.

obnb.metric.standard.auroc(y_true, y_predict)[source]

AUROC metric.

Return type:

float

Parameters:

y_true (ndarray) –
y_predict (ndarray) –

obnb.metric.standard.cast_ndarray_type(x)[source]

Cast numpy ndarray type.

Return type:: ndarray
Parameters:: x (ndarray | Tensor) –

obnb.metric.standard.log2_auprc_prior(y_true, y_predict)[source]

Log2 auprc over prior.

Return type:

float

Parameters:

y_true (ndarray) –
y_predict (ndarray) –

obnb.metric.standard.precision_at_topk(y_true, y_predict)[source]

Precision at top k.

Return type:

float

Parameters:

y_true (ndarray) –
y_predict (ndarray) –

obnb.metric.standard.prior(y_true)[source]

Return the prior of a label vector.

The portion of positive examples.

Return type:: float
Parameters:: y_true (ndarray) –

obnb.metric.standard.wrap_metric(metric_func)[source]

Wrap metric function with common processing steps.

Skip computation if None
Perturn reduction when calculating metrics in a multi-class setting

GrpahGym patches

Custom metrics compatible with GraphGym logger.

obnb.metric.graphgym_metric.graphgym_auroc(y_true, y_predict)

AUROC metric.

Return type:

float

Parameters:

y_true (ndarray) –
y_predict (ndarray) –

obnb.metric.graphgym_metric.graphgym_log2_auprc_prior(y_true, y_predict)

Log2 auprc over prior.

Return type:

float

Parameters:

y_true (ndarray) –
y_predict (ndarray) –

obnb.metric.graphgym_metric.graphgym_precision_at_topk(y_true, y_predict)

Precision at top k.

Return type:

float

Parameters:

y_true (ndarray) –
y_predict (ndarray) –