23ws-LLMcoder/LLMcoder-GitHub-Python-Mix-Direct

1 2import numpy as np3import matplotlib.pyplot as plt4from sklearn.tree import DecisionTreeClassifier5from ipywidgets import interact6 7 8def visualize_tree(estimator, X, y, boundaries=True,9                   xlim=None, ylim=None, ax=None):10    ax = ax or plt.gca()11    12    # Plot the training points13    ax.scatter(X[:, 0], X[:, 1], c=y, s=30, cmap='viridis',14               clim=(y.min(), y.max()), zorder=3)15    ax.axis('tight')16    ax.axis('off')17    if xlim is None:18        xlim = ax.get_xlim()19    if ylim is None:20        ylim = ax.get_ylim()21    22    # fit the estimator23    estimator.fit(X, y)24    xx, yy = np.meshgrid(np.linspace(*xlim, num=200),25                         np.linspace(*ylim, num=200))26    Z = estimator.predict(np.c_[xx.ravel(), yy.ravel()])27 28    # Put the result into a color plot29    n_classes = len(np.unique(y))30    Z = Z.reshape(xx.shape)31    contours = ax.contourf(xx, yy, Z, alpha=0.3,32                           levels=np.arange(n_classes + 1) - 0.5,33                           cmap='viridis', clim=(y.min(), y.max()),34                           zorder=1)35 36    ax.set(xlim=xlim, ylim=ylim)37    38    # Plot the decision boundaries39    def plot_boundaries(i, xlim, ylim):40        if i >= 0:41            tree = estimator.tree_42        43            if tree.feature[i] == 0:44                ax.plot([tree.threshold[i], tree.threshold[i]], ylim, '-k', zorder=2)45                plot_boundaries(tree.children_left[i],46                                [xlim[0], tree.threshold[i]], ylim)47                plot_boundaries(tree.children_right[i],48                                [tree.threshold[i], xlim[1]], ylim)49        50            elif tree.feature[i] == 1:51                ax.plot(xlim, [tree.threshold[i], tree.threshold[i]], '-k', zorder=2)52                plot_boundaries(tree.children_left[i], xlim,53                                [ylim[0], tree.threshold[i]])54                plot_boundaries(tree.children_right[i], xlim,55                                [tree.threshold[i], ylim[1]])56            57    if boundaries:58        plot_boundaries(0, xlim, ylim)59 60 61def plot_tree_interactive(X, y):62    def interactive_tree(depth=5):63        clf = DecisionTreeClassifier(max_depth=depth, random_state=0)64        visualize_tree(clf, X, y)65 66    return interact(interactive_tree, depth=[1, 5])67 68 69def randomized_tree_interactive(X, y):70    N = int(0.75 * X.shape[0])71    72    xlim = (X[:, 0].min(), X[:, 0].max())73    ylim = (X[:, 1].min(), X[:, 1].max())74    75    def fit_randomized_tree(random_state=0):76        clf = DecisionTreeClassifier(max_depth=15)77