如何将图例 colors 与 plot colors 匹配？

Question

I'm working with a pretty simple example. I create three scatter plots on the same set of axes, and each data set I plot has a different associated colormap. However, the legend does not look as I'd want it to; why is this?

import matplotlib.pyplot as plt
from sklearn.datasets import make_blobs

X, y = make_blobs(n_samples=500, n_features=2, cluster_std=1.0, centers=[(0,0), (3,3)])

plt.scatter(X[:, 0], X[:, 1], c=y, s=50, cmap='rainbow');


clf_tree = DecisionTreeClassifier(criterion='entropy', max_depth=1)
clf_tree = clf_tree.fit(X, y)
y_pred = clf_tree.predict(X)

clf_tree = DecisionTreeClassifier(criterion='entropy', min_samples_leaf = 3)
clf_tree = clf_tree.fit(X, y)
y_pred = clf_tree.predict(X)

clf_tree = DecisionTreeClassifier(criterion='entropy', max_leaf_nodes = 3)
clf_tree = clf_tree.fit(X, y)
y_pred = clf_tree.predict(X)



#shap.decision_plot(expected_value, sh, features_display, link='logit', highlight=misclassified)

for i in range(len(y)):
    if y[i] != y_pred[i]:
        plt.scatter(X[i, 0], X[i, 1], c=y[i], s=50, cmap='Dark2')
plt.legend(("0","1","miss"))

Answer 1

You can set them manually like this:

plt.legend(("0","1","miss"))
ax = plt.gca()
leg = ax.get_legend()
leg.legendHandles[0].set_color('purple')
leg.legendHandles[1].set_color('red')
leg.legendHandles[2].set_color('green')
plt.show()