In [1]:

    

from sklearn import cross_validation


    

In [2]:

    

from sklearn import datasets


    

In [3]:

    

iris = datasets.load_iris()


    

In [21]:

    

data_train, data_test, target_train, target_test = cross_validation.train_test_split(iris.data[:, :2], 
                                                                                     iris.target, test_size=0.3333, random_state = 0)


    

In [14]:

    

from sklearn import tree


    

In [23]:

    

clf = tree.DecisionTreeClassifier()


    

In [24]:

    

clf = clf.fit(data_train, target_train)


    

In [25]:

    

pred = clf.predict(data_test)


    

In [26]:

    

1.0*sum(pred!=target_test)/len(target_test)


    

    
Out[26]:





0.32000000000000001

In [27]:

    

clf.score(data_test, target_test)


    

    
Out[27]:





0.68000000000000005

In [28]:

    

scores = cross_validation.cross_val_score(clf, data_train, target_train, cv=5)


    

In [29]:

    

print scores


    

    




[ 0.8   0.7   0.7   0.5   0.75]

In [30]:

    

print scores.mean()


    

    




0.69

In [48]:

    

bestPruning = -1
bestAccuracy = 0
for i in range(2, len(target_train)):
    clf = tree.DecisionTreeClassifier()
    scores = cross_validation.cross_val_score(clf, data_train, target_train, cv=i)
    if scores.mean() > bestAccuracy:
        bestAccuracy = scores.mean()
        bestPruning = i


    

In [49]:

    

print bestAccuracy
print bestPruning


    

    




0.789661319073
3

In [47]:

    




    

In [ ]: