Skip to content

plot_cv_splits

method plot_cv_splits(models=None, title=None, legend="upper right", figsize=(900, 600), filename=None, display=True)[source]

Visualize the cross-validation splits.

Plots the train and test splits for each cross-validation iteration of the cv object. The x-axis shows the number of rows, where every point corresponds to the n-th sample. The top bar shows the original train/test split. Additionally, class labels and groups are plotted when relevant.

Warning

This plot is only available for models that ran cross-validation using the cross_validate method.

Parameters models: int, str, Model, segment, sequence or None, default=None
Model to plot. If None, all models are selected. Note that leaving the default option could raise an exception if there are multiple models. To avoid this, call the plot directly from a model, e.g., atom.lr.plot_cv_splits().

title: str, dict or None, default=None
Title for the plot.

legend: str, dict or None, default="upper right"
Legend for the plot. See the user guide for an extended description of the choices.

  • If None: No legend is shown.
  • If str: Position to display the legend.
  • If dict: Legend configuration.

figsize: tuple, default=(900, 600)
Figure's size in pixels, format as (x, y).

filename: str, Path or None, default=None
Save the plot using this name. Use "auto" for automatic naming. The type of the file depends on the provided name (.html, .png, .pdf, etc...). If filename has no file type, the plot is saved as html. If None, the plot is not saved.

display: bool or None, default=True
Whether to render the plot. If None, it returns the figure.

Returns{#plot_cv_splits-go.Figure or None} go.Figure or None
Plot object. Only returned if display=None.


See Also

plot_data_splits

Visualize the data splits.

plot_decomposition

Plot the trend, seasonality and residuals of a time series.

plot_relationships

Plot pairwise relationships in a dataset.


Example

>>> from atom import ATOMClassifier, ATOMForecaster
>>> from random import choices
>>> from sklearn.datasets import load_breast_cancer
>>> from sktime.datasets import load_airline

>>> X, y = load_breast_cancer(return_X_y=True, as_frame=True)

>>> # Without groups
>>> atom = ATOMClassifier(X, y, shuffle=False, n_rows=0.2, random_state=1)
>>> atom.run("LR", metric=["f1", "auc"])
>>> atom.lr.cross_validate(cv=4)
  train_f1 test_f1 train_auc test_auc time
0 0.983051 0.962963 1.000000 1.000000 0.021019
1 0.991150 0.937500 1.000000 0.994872 0.021019
2 0.980000 0.976744 1.000000 1.000000 0.021019
3 0.990099 0.976744 0.999429 0.993197 0.022020
mean 0.986075 0.963488 0.999857 0.997017 0.021270 

>>> atom.plot_cv_splits()


>>> # With groups
>>> groups = choices(["A", "B", "C", "D"], k=X.shape[0])
>>> atom = ATOMClassifier(X, y, metadata={"groups": groups}, n_rows=0.2, random_state=1)
>>> atom.run("LR", metric=["f1", "auc"])
>>> atom.lr.cross_validate(cv=4)
  train_f1 test_f1 train_auc test_auc time
0 1.000000 0.947368 1.000000 0.981481 0.022021
1 0.990476 0.972973 1.000000 1.000000 0.021019
2 0.981481 0.941176 1.000000 1.000000 0.021019
3 0.982143 0.937500 0.999449 0.993056 0.021019
mean 0.988525 0.949754 0.999862 0.993634 0.021270 

>>> atom.plot_cv_splits()


>>> # For forecast models
>>> y = load_airline()

>>> atom = ATOMForecaster(y, random_state=1)
>>> atom.run("Croston", metric=["mape", "mse", "mae"])
>>> atom.croston.cross_validate(cv=4)
  test_mape test_mse test_mae time
0 0.263562 4248.686875 57.152766 0.001885
1 0.170446 3539.682857 47.622303 0.001879
2 0.218499 10212.231474 85.077206 0.001953
3 0.144202 8875.839953 69.594088 0.001915
mean 0.199177 6719.110290 64.861591 0.001908 

>>> atom.plot_cv_splits()