MAINT Use class_of_interest in DecisionBoundaryDisplay

[ArturoAmorQ]

Since scikit-learn v1.4 DecisionBoundaryDisplay accepts class_of_interest for multiclass visualization. This feature is promised in the current version of the MOOC.

Notice that, as it requires updating the minimal version, it may change the experience of current enrolled participants.

[glemaitre]

I see an occurrence where we could reuse the axis from the first display, it seems more natural.

import matplotlib.pyplot as plt
import matplotlib as mpl
import seaborn as sns

from sklearn.inspection import DecisionBoundaryDisplay

tab10_norm = mpl.colors.Normalize(vmin=-0.5, vmax=8.5)

palette = ["tab:blue", "tab:green", "tab:orange"]
disp = DecisionBoundaryDisplay.from_estimator(
    tree,
    data_train,
    response_method="predict",
    cmap="tab10",
    norm=tab10_norm,
    alpha=0.5,
)
sns.scatterplot(
    data=penguins,
    x=culmen_columns[0],
    y=culmen_columns[1],
    hue=target_column,
    palette=palette,
    ax=disp.ax_,
)
disp.ax_.legend(bbox_to_anchor=(1.05, 1), loc="upper left")
_ = disp.ax_.set_title("Decision boundary using a decision tree")

[glemaitre]

LGTM otherwise.

[ogrisel]

If this is already there in 1.4, we can directly target main, no?

[ArturoAmorQ]

If this is already there in 1.4, we can directly target main, no?

According to #789, we are currently running on v1.3. I confirmed this by running sklearn.__version__ on the mooc.

[ogrisel]

I did another pass of rephrasing.

+1 for merging to update-to-scikit-learn-1.6 and pushing a resync of the notebooks / exercises in that branch to check the CI.