tfma.metrics.AUC

Approximates the AUC (Area under the curve) of the ROC or PR curves.

Inherits From: Metric

The AUC (Area under the curve) of the ROC (Receiver operating characteristic; default) or PR (Precision Recall) curves are quality measures of binary classifiers. Unlike the accuracy, and like cross-entropy losses, ROC-AUC and PR-AUC evaluate all the operational points of a model.

This class approximates AUCs using a Riemann sum. During the metric accumulation phase, predictions are accumulated within predefined buckets by value. The AUC is then computed by interpolating per-bucket averages. These buckets define the evaluated operational points.

This metric uses true_positives, true_negatives, false_positives and false_negatives to compute the AUC. To discretize the AUC curve, a linearly spaced set of thresholds is used to compute pairs of recall and precision values. The area under the ROC-curve is therefore computed using the height of the recall values by the false positive rate, while the area under the PR-curve is the computed using the height of the precision values by the recall.

This value is ultimately returned as auc, an idempotent operation that computes the area under a discretized curve of precision versus recall values (computed using the aforementioned variables). The num_thresholds variable controls the degree of discretization with larger numbers of thresholds more closely approximating the true AUC. The quality of the approximation may vary dramatically depending on num_thresholds. The thresholds parameter can be used to manually specify thresholds which split the predictions more evenly.

For a best approximation of the real AUC, predictions should be distributed approximately uniformly in the range [0, 1]. The quality of the AUC approximation may be poor if this is not the case. Setting summation_method to 'minoring' or 'majoring' can help quantify the error in the approximation by providing lower or upper bound estimate of the AUC.

If sample_weight is None, weights default to 1. Use sample_weight of 0 to mask values.

num_thresholds (Optional) Defaults to 10000. The number of thresholds to use when discretizing the roc curve. Values must be > 1.
curve (Optional) Specifies the name of the curve to be computed, 'ROC' [default] or 'PR' for the Precision-Recall-curve.
summation_method (Optional) Specifies the Riemann summation method used. 'interpolation' (default) applies mid-point summation scheme for ROC. For PR-AUC, interpolates (true/false) positives but not the ratio that is precision (see Davis & Goadrich 2006 for details); 'minoring' applies left summation for increasing intervals and right summation for decreasing intervals; 'majoring' does the opposite.
name (Optional) string name of the metric instance.
thresholds (Optional) A list of floating point values to use as the thresholds for discretizing the curve. If set, the num_thresholds parameter is ignored. Values should be in [0, 1]. Endpoint thresholds equal to {-epsilon, 1+epsilon} for a small positive epsilon value will be automatically included with these to correctly handle predictions equal to exactly 0 or 1.
top_k (Optional) Used with a multi-class model to specify that the top-k values should be used to compute the confusion matrix. The net effect is that the non-top-k values are set to -inf and the matrix is then constructed from the average TP, FP, TN, FN across the classes. When top_k is used, metrics_specs.binarize settings must not be present. Only one of class_id or top_k should be configured. When top_k is set, the default thresholds are [float('-inf')].
class_id (Optional) Used with a multi-class model to specify which class to compute the confusion matrix for. When class_id is used, metrics_specs.binarize settings must not be present. Only one of class_id or top_k should be configured.

compute_confidence_interval Whether to compute confidence intervals for this metric.

Note that this may not completely remove the computational overhead involved in computing a given metric. This is only respected by the jackknife confidence interval method.

Methods

computations

View source

Creates computations associated with metric.

from_config

View source

get_config

View source

Returns serializable config.