.. DO NOT EDIT. .. THIS FILE WAS AUTOMATICALLY GENERATED BY SPHINX-GALLERY. .. TO MAKE CHANGES, EDIT THE SOURCE PYTHON FILE: .. "auto_examples/plot_leave_neighbors.py" .. LINE NUMBERS ARE GIVEN BELOW. .. only:: html .. note:: :class: sphx-glr-download-link-note :ref:`Go to the end ` to download the full example code. .. rst-class:: sphx-glr-example-title .. _sphx_glr_auto_examples_plot_leave_neighbors.py: Close leaves in a decision trees ================================ A decision tree computes a partition of the feature space. We can wonder which leave is close to another one even though the predict the same value (or class). Do they share a border? .. GENERATED FROM PYTHON SOURCE LINES 11-13 A simple tree +++++++++++++ .. GENERATED FROM PYTHON SOURCE LINES 13-36 .. code-block:: Python import matplotlib.pyplot as plt import numpy from mlinsights.mltree import predict_leaves, tree_leave_index, tree_leave_neighbors from sklearn.datasets import load_iris from sklearn.tree import DecisionTreeClassifier X = numpy.array( [[10, 0], [10, 1], [10, 2], [11, 0], [11, 1], [11, 2], [12, 0], [12, 1], [12, 2]] ) y = list(range(X.shape[0])) # In[5]: fig, ax = plt.subplots(1, 1) for i in range(X.shape[0]): ax.plot([X[i, 0]], [X[i, 1]], "o", ms=19, label="y=%d" % y[i]) ax.legend() ax.set_title("Simple grid") .. image-sg:: /auto_examples/images/sphx_glr_plot_leave_neighbors_001.png :alt: Simple grid :srcset: /auto_examples/images/sphx_glr_plot_leave_neighbors_001.png :class: sphx-glr-single-img .. rst-class:: sphx-glr-script-out .. code-block:: none Text(0.5, 1.0, 'Simple grid') .. GENERATED FROM PYTHON SOURCE LINES 38-43 .. code-block:: Python clr = DecisionTreeClassifier(max_depth=5) clr.fit(X, y) .. raw:: html

DecisionTreeClassifier(max_depth=5)

In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook.
On GitHub, the HTML representation is unable to render, please try loading this page with nbviewer.org.

.. GENERATED FROM PYTHON SOURCE LINES 44-45 The contains the following list of leaves. .. GENERATED FROM PYTHON SOURCE LINES 45-50 .. code-block:: Python tree_leave_index(clr) .. rst-class:: sphx-glr-script-out .. code-block:: none [2, 4, 5, 8, 9, 12, 13, 15, 16] .. GENERATED FROM PYTHON SOURCE LINES 51-52 Let's compute the neighbors for each leave. .. GENERATED FROM PYTHON SOURCE LINES 52-57 .. code-block:: Python neighbors = tree_leave_neighbors(clr) neighbors .. rst-class:: sphx-glr-script-out .. code-block:: none {(np.int32(2), np.int32(4)): [(np.int64(0), (np.float64(10.0), np.float64(0.0)), (np.float64(11.0), np.float64(0.0)))], (np.int32(2), np.int32(8)): [(np.int64(1), (np.float64(10.0), np.float64(0.0)), (np.float64(10.0), np.float64(1.0)))], (np.int32(8), np.int32(12)): [(np.int64(0), (np.float64(10.0), np.float64(1.0)), (np.float64(11.0), np.float64(1.0)))], (np.int32(8), np.int32(9)): [(np.int64(1), (np.float64(10.0), np.float64(1.0)), (np.float64(10.0), np.float64(2.0)))], (np.int32(9), np.int32(13)): [(np.int64(0), (np.float64(10.0), np.float64(2.0)), (np.float64(11.0), np.float64(2.0)))], (np.int32(4), np.int32(5)): [(np.int64(0), (np.float64(11.0), np.float64(0.0)), (np.float64(12.0), np.float64(0.0)))], (np.int32(4), np.int32(12)): [(np.int64(1), (np.float64(11.0), np.float64(0.0)), (np.float64(11.0), np.float64(1.0)))], (np.int32(12), np.int32(15)): [(np.int64(0), (np.float64(11.0), np.float64(1.0)), (np.float64(12.0), np.float64(1.0)))], (np.int32(12), np.int32(13)): [(np.int64(1), (np.float64(11.0), np.float64(1.0)), (np.float64(11.0), np.float64(2.0)))], (np.int32(13), np.int32(16)): [(np.int64(0), (np.float64(11.0), np.float64(2.0)), (np.float64(12.0), np.float64(2.0)))], (np.int32(5), np.int32(15)): [(np.int64(1), (np.float64(12.0), np.float64(0.0)), (np.float64(12.0), np.float64(1.0)))], (np.int32(15), np.int32(16)): [(np.int64(1), (np.float64(12.0), np.float64(1.0)), (np.float64(12.0), np.float64(2.0)))]} .. GENERATED FROM PYTHON SOURCE LINES 58-59 And let's explain the results by drawing the segments ``[x1, x2]``. .. GENERATED FROM PYTHON SOURCE LINES 59-82 .. code-block:: Python leaves = predict_leaves(clr, X) fig, ax = plt.subplots(1, 2, figsize=(14, 4)) for i in range(X.shape[0]): ax[0].plot([X[i, 0]], [X[i, 1]], "o", ms=19) ax[1].plot([X[i, 0]], [X[i, 1]], "o", ms=19) ax[0].text(X[i, 0] + 0.1, X[i, 1] - 0.1, "y=%d\nl=%d" % (y[i], leaves[i])) for edge, segments in neighbors.items(): for segment in segments: # leaves l1, l2 are neighbors l1, l2 = edge # the common border is [x1, x2] x1 = segment[1] x2 = segment[2] ax[1].plot([x1[0], x2[0]], [x1[1], x2[1]], "b--") ax[1].text((x1[0] + x2[0]) / 2, (x1[1] + x2[1]) / 2, "%d->%d" % edge) ax[0].set_title("Classes and leaves") ax[1].set_title("Segments") .. image-sg:: /auto_examples/images/sphx_glr_plot_leave_neighbors_002.png :alt: Classes and leaves, Segments :srcset: /auto_examples/images/sphx_glr_plot_leave_neighbors_002.png :class: sphx-glr-single-img .. rst-class:: sphx-glr-script-out .. code-block:: none Text(0.5, 1.0, 'Segments') .. GENERATED FROM PYTHON SOURCE LINES 83-85 On Iris +++++++ .. GENERATED FROM PYTHON SOURCE LINES 85-90 .. code-block:: Python iris = load_iris() .. GENERATED FROM PYTHON SOURCE LINES 92-98 .. code-block:: Python X = iris.data[:, :2] y = iris.target .. GENERATED FROM PYTHON SOURCE LINES 100-106 .. code-block:: Python clr = DecisionTreeClassifier(max_depth=3) clr.fit(X, y) .. raw:: html

DecisionTreeClassifier(max_depth=3)

.. GENERATED FROM PYTHON SOURCE LINES 108-153 .. code-block:: Python def draw_border( clr, X, y, fct=None, incx=1, incy=1, figsize=None, border=True, ax=None ): # see https://sashat.me/2017/01/11/list-of-20-simple-distinct-colors/ # https://matplotlib.org/examples/color/colormaps_reference.html h = 0.02 # step size in the mesh # Plot the decision boundary. For that, we will assign a color to each # point in the mesh [x_min, x_max]x[y_min, y_max]. x_min, x_max = X[:, 0].min() - incx, X[:, 0].max() + incx y_min, y_max = X[:, 1].min() - incy, X[:, 1].max() + incy xx, yy = numpy.meshgrid( numpy.arange(x_min, x_max, h), numpy.arange(y_min, y_max, h) ) if fct is None: Z = clr.predict(numpy.c_[xx.ravel(), yy.ravel()]) else: Z = fct(clr, numpy.c_[xx.ravel(), yy.ravel()]) # Put the result into a color plot cmap = plt.cm.tab20 Z = Z.reshape(xx.shape) if ax is None: fig, ax = plt.subplots(1, 1, figsize=figsize or (4, 3)) ax.pcolormesh(xx, yy, Z, cmap=cmap) # Plot also the training points ax.scatter(X[:, 0], X[:, 1], c=y, edgecolors="k", cmap=cmap) ax.set_xlabel("Sepal length") ax.set_ylabel("Sepal width") ax.set_xlim(xx.min(), xx.max()) ax.set_ylim(yy.min(), yy.max()) return ax fig, ax = plt.subplots(1, 2, figsize=(14, 4)) draw_border(clr, X, y, border=False, ax=ax[0]) ax[0].set_title("Iris") draw_border(clr, X, y, border=False, ax=ax[1], fct=lambda m, x: predict_leaves(m, x)) ax[1].set_title("Leaves") .. image-sg:: /auto_examples/images/sphx_glr_plot_leave_neighbors_003.png :alt: Iris, Leaves :srcset: /auto_examples/images/sphx_glr_plot_leave_neighbors_003.png :class: sphx-glr-single-img .. rst-class:: sphx-glr-script-out .. code-block:: none Text(0.5, 1.0, 'Leaves') .. GENERATED FROM PYTHON SOURCE LINES 155-161 .. code-block:: Python neighbors = tree_leave_neighbors(clr) list(neighbors.items())[:2] .. rst-class:: sphx-glr-script-out .. code-block:: none [((np.int32(3), np.int32(4)), [(np.int64(0), (np.float64(4.650000095367432), np.float64(2.4750000834465027)), (np.float64(5.025000095367432), np.float64(2.4750000834465027)))]), ((np.int32(3), np.int32(6)), [(np.int64(1), (np.float64(4.650000095367432), np.float64(2.4750000834465027)), (np.float64(4.650000095367432), np.float64(3.1250000596046448)))])] .. GENERATED FROM PYTHON SOURCE LINES 163-188 .. code-block:: Python fig, ax = plt.subplots(1, 1, figsize=(8, 8)) draw_border( clr, X, y, incx=1, incy=1, figsize=(6, 4), border=False, ax=ax, fct=lambda m, x: predict_leaves(m, x), ) for edge, segments in neighbors.items(): for segment in segments: # leaves l1, l2 are neighbors l1, l2 = edge # the common border is [x1, x2] x1 = segment[1] x2 = segment[2] ax.plot([x1[0], x2[0]], [x1[1], x2[1]], "b--") ax.text((x1[0] + x2[0]) / 2, (x1[1] + x2[1]) / 2, "%d->%d" % edge) ax.set_title("Leaves and segments") .. image-sg:: /auto_examples/images/sphx_glr_plot_leave_neighbors_004.png :alt: Leaves and segments :srcset: /auto_examples/images/sphx_glr_plot_leave_neighbors_004.png :class: sphx-glr-single-img .. rst-class:: sphx-glr-script-out .. code-block:: none Text(0.5, 1.0, 'Leaves and segments') .. rst-class:: sphx-glr-timing **Total running time of the script:** (0 minutes 1.007 seconds) .. _sphx_glr_download_auto_examples_plot_leave_neighbors.py: .. only:: html .. container:: sphx-glr-footer sphx-glr-footer-example .. container:: sphx-glr-download sphx-glr-download-jupyter :download:`Download Jupyter notebook: plot_leave_neighbors.ipynb ` .. container:: sphx-glr-download sphx-glr-download-python :download:`Download Python source code: plot_leave_neighbors.py ` .. container:: sphx-glr-download sphx-glr-download-zip :download:`Download zipped: plot_leave_neighbors.zip ` .. only:: html .. rst-class:: sphx-glr-signature `Gallery generated by Sphinx-Gallery `_