napistu_torch.visualization.basic_metrics

Basic metrics like train loss and test/val AUC.

Functions

plot_auc_only(summaries, display_names[, ...])

Plot only test/val AUC comparison.

plot_model_comparison(summaries, display_names)

Create comparison plots for model training loss and test/val AUC.
napistu_torch.visualization.basic_metrics._extract_metric(summaries: Dict[str, Dict[str, Any]], metric_key: str) → List[float | None]

Extract a metric from all model summaries.

napistu_torch.visualization.basic_metrics._plot_test_val_auc(ax: Axes, display_names: List[str], test_aucs: List[float], val_aucs: List[float], ylim: Tuple[float, float] | None = None, bar_width: float = 0.35, title: str | None = None, horizontal: bool = False) → None

Plot test and validation AUC as grouped bar chart.

Parameters:

  • ax (plt.Axes) – Matplotlib axes object to plot on

  • display_names (List[str]) – Model names for axis labels

  • test_aucs (List[float]) – Test AUC values for each model

  • val_aucs (List[float]) – Validation AUC values for each model

  • ylim (Optional[Tuple[float, float]]) – Axis limits for AUC values as (min, max). If None, calculated automatically

  • bar_width (float) – Width of each bar in the grouped bar chart

  • title (Optional[str]) – Custom title for the plot. If None, uses default title

  • horizontal (bool) – If True, create horizontal bars. If False (default), create vertical bars

napistu_torch.visualization.basic_metrics._plot_train_loss(ax: Axes, display_names: List[str], train_losses: List[float], ylim: Tuple[float, float] | None = None) → None

Plot training loss as a bar chart.

Parameters:

  • ax (plt.Axes) – Matplotlib axes object to plot on

  • display_names (List[str]) – Model names for x-axis labels

  • train_losses (List[float]) – Training loss values for each model

  • ylim (Optional[Tuple[float, float]]) – Y-axis limits as (min, max). If None, calculated automatically

napistu_torch.visualization.basic_metrics.plot_auc_only(summaries: Dict[str, Dict[str, Any]], display_names: List[str], figsize: Tuple[int, int] = (10, 6), test_auc_attribute: str = 'test_auc', val_auc_attribute: str = 'val_auc', title: str | None = None, horizontal: bool = True, ax: Axes | None = None, **kwargs) → Tuple[Figure, Axes]

Plot only test/val AUC comparison.

Parameters:

  • summaries (Dict[str, Dict[str, Any]]) – Dictionary mapping model names to their summary metrics. Each summary must contain ‘test_auc’ and ‘val_auc’.

  • display_names (List[str]) – Clean display names for models (must match order of summaries.keys())

  • figsize (Tuple[int, int]) – Figure size as (width, height)

  • test_auc_attribute (str) – Attribute name for test AUC in summaries

  • val_auc_attribute (str) – Attribute name for validation AUC in summaries

  • title (Optional[str]) – Custom title for the plot. If None, uses default title

  • horizontal (bool) – If True, create horizontal bars. If False (default), create vertical bars

  • ax (Optional[plt.Axes]) – Matplotlib axes object to plot on. If None, creates a new figure and axes.

  • **kwargs (dict) – Additional keyword arguments to pass to _plot_test_val_auc

Returns:

Figure and axis objects

Return type:

Tuple[plt.Figure, plt.Axes]

Examples

>>> summaries = {
...     'model1': {'test_auc': 0.75, 'val_auc': 0.74},
...     'model2': {'test_auc': 0.78, 'val_auc': 0.77}
... }
>>> display_names = ['Model 1', 'Model 2']
>>> fig, ax = plot_auc_only(summaries, display_names)
>>> plt.show()
napistu_torch.visualization.basic_metrics.plot_model_comparison(summaries: Dict[str, Dict[str, Any]], display_names: List[str], figsize: Tuple[int, int] = (16, 6), train_loss_attribute: str = 'train_loss', test_auc_attribute: str = 'test_auc', val_auc_attribute: str = 'val_auc') → Tuple[Figure, Tuple[Axes, Axes]]

Create comparison plots for model training loss and test/val AUC.

Parameters:

  • summaries (Dict[str, Dict[str, Any]]) – Dictionary mapping model names to their summary metrics. Each summary must contain ‘train_loss’, ‘test_auc’, and ‘val_auc’.

  • display_names (List[str]) – Clean display names for models (must match order of summaries.keys())

  • figsize (Tuple[int, int]) – Figure size as (width, height)

  • train_loss_attribute (str) – Attribute name for train loss in summaries

  • test_auc_attribute (str) – Attribute name for test AUC in summaries

  • val_auc_attribute (str) – Attribute name for validation AUC in summaries

Returns:

Figure and axes tuple (ax1 for train loss, ax2 for AUC)

Return type:

Tuple[plt.Figure, Tuple[plt.Axes, plt.Axes]]

Examples

>>> summaries = {
...     'model1': {'train_loss': 1.5, 'test_auc': 0.75, 'val_auc': 0.74},
...     'model2': {'train_loss': 1.2, 'test_auc': 0.78, 'val_auc': 0.77}
... }
>>> display_names = ['Model 1', 'Model 2']
>>> fig, (ax1, ax2) = plot_model_comparison(summaries, display_names)
>>> plt.show()