matcha.utils.plotting

Plotting utilities for regression and classification model evaluation.

Provides interactive Plotly-based visualizations including scatter plots with trendlines and fold-error boundaries for regression, and ROC/PR curves with probability histograms for classification.

Functions

plot_regression(true_values, pred_values[, ...])

Create a scatter plot of true vs predicted values with trendline using Plotly.

plot_classification(true_values, pred_values, prob_values)

Create a three-panel classification plot with ROC, PR, and probability histograms.

save_plot(fig, file_path)

Save a Plotly figure as an HTML file.

Module Contents

matcha.utils.plotting.plot_regression(true_values: numpy.ndarray, pred_values: numpy.ndarray, plot_title: str = 'Regression performance', labels: list = None, is_log10: bool = False)[source]

Create a scatter plot of true vs predicted values with trendline using Plotly.

Includes OLS trendline, perfect prediction diagonal, and 2-fold/3-fold error boundary lines.

Parameters:

  • true_values (numpy.ndarray) – Array of true target values.

  • pred_values (numpy.ndarray) – Array of predicted target values.

  • plot_title (str) – Title for the plot.

  • labels (list) – Optional list of labels for hover data corresponding to each data point.

  • is_log10 (bool) – Whether the values are log10-transformed. Adjusts fold-error boundary calculations accordingly.

Returns:

A Plotly figure object, or None if no valid data points exist.

Return type:

plotly.graph_objects.Figure or None

matcha.utils.plotting.plot_classification(true_values: numpy.ndarray, pred_values: numpy.ndarray, prob_values: numpy.ndarray, plot_title: str = 'Classification performance', model: matcha.sklearn.base_sklearn_model.BaseScikitLearnModel | None = None)[source]

Create a three-panel classification plot with ROC, PR, and probability histograms.

Generates an interactive Plotly figure with ROC-AUC curve, Precision-Recall curve, and predicted probability distribution histograms for each class.

Parameters:

  • true_values (numpy.ndarray) – Array of true binary labels (0 or 1).

  • pred_values (numpy.ndarray) – Array of predicted binary labels (0 or 1).

  • prob_values (numpy.ndarray) – Array of predicted probabilities for the positive class.

  • plot_title (str) – Title for the plot.

  • model (BaseScikitLearnModel or None) – Optional model instance used to encode labels via its internal label encoder.

Returns:

A Plotly figure object, or None if no valid data points exist.

Return type:

plotly.graph_objects.Figure or None

matcha.utils.plotting.save_plot(fig, file_path)[source]

Save a Plotly figure as an HTML file.

Creates parent directories if they do not exist.

Parameters:

  • fig (plotly.graph_objects.Figure) – The Plotly figure to save.

  • file_path (str) – Destination file path (should end in .html).