scikit-learn/examples/tree/plot_tree_regression.py

"""
===================================================================
Decision Tree Regression
===================================================================

1D regression with :ref:`decision trees <tree>`: the decision tree is
used to fit a sine curve with addition noisy observation. As a result, it
learn local linear regressions approximating the sine curve.

We can see that if the maximum depth of the tree (controled by the
`max_depth` parameter) is set to high, the decision trees learn too fine
details of the training data and learn from the noise, i.e. they overfit.
"""
print __doc__

###############################################################################
# Generate sample data
import numpy as np

# Create a random number generator
rng = np.random.RandomState(1)
X = np.sort(5*rng.rand(80, 1), axis=0)
y = np.sin(X).ravel()

###############################################################################
# Add noise to targets
y[::5] += 3*(0.5 - rng.rand(16))

###############################################################################
# Fit regression model
from sklearn.tree import DecisionTreeRegressor

clf_1 = DecisionTreeRegressor(max_depth=2)
clf_2 = DecisionTreeRegressor(max_depth=5)
y_1 = clf_1.fit(X, y).predict(X)
y_2 = clf_2.fit(X, y).predict(X)

###############################################################################
# look at the results
import pylab as pl
pl.figure(1, figsize=(5, 4))
pl.clf()
pl.scatter(X, y, c='k', label='data')
pl.plot(X, y_1, c='g', label='max_depth=2', linewidth=2)
pl.plot(X, y_2, c='r', label='max_depth=5', linewidth=2)
pl.axis('tight')
pl.xlabel('data')
pl.ylabel('target')
pl.title('Decision Tree Regression')
pl.legend(loc='best')
pl.show()