evaluation.metrics
evaluation.metrics
Metrics.py provides a set of evaluation metrics for multiclass, single-label classification tasks.
Each metric is designed to assess the performance of a classification model in different ways, providing insights into its accuracy, precision, recall, and overall balance.
Classes
| Name | Description |
|---|---|
| ClassificationAccuracy | Calculate the accuracy of results. |
| ClassificationMacroF1 | Calculate macro F1 by averaging per-label F1. |
| ClassificationMacroPrecision | Calculate macro precision by averaging per-label precision. |
| ClassificationMacroRecall | Calculate macro recall by averaging per-label recall. |
| Metric | Base class for all classification metrics. |
| MetricResult | Represents the result of a metric evaluation. |
ClassificationAccuracy
evaluation.metrics.ClassificationAccuracy()Calculate the accuracy of results.
ClassificationMacroF1
evaluation.metrics.ClassificationMacroF1()Calculate macro F1 by averaging per-label F1.
ClassificationMacroPrecision
evaluation.metrics.ClassificationMacroPrecision()Calculate macro precision by averaging per-label precision.
ClassificationMacroRecall
evaluation.metrics.ClassificationMacroRecall()Calculate macro recall by averaging per-label recall.
Metric
evaluation.metrics.Metric()Base class for all classification metrics.
Methods
| Name | Description |
|---|---|
| evaluate | Evaluate the metric on the provided evaluation data. |
evaluate
evaluation.metrics.Metric.evaluate(eval_data)Evaluate the metric on the provided evaluation data.
Parameters
| Name | Type | Description | Default |
|---|---|---|---|
| eval_data | pd.DataFrame | DataFrame with ‘doc_label’ and ‘ground_truth_label’ columns. | required |
Returns
| Name | Type | Description |
|---|---|---|
| MetricResult | MetricResult containing the metric name and computed value. |
MetricResult
evaluation.metrics.MetricResult(name, value)Represents the result of a metric evaluation.