rcParams['figure.figsize'] = (16, 4)

# A lot of code below borrowed from the scikits learn docs

from sklearn.datasets import load_iris
data = load_iris()

from sklearn import tree

clf = tree.DecisionTreeClassifier()
clf.fit(data.data, data.target)

clf.predict(data.data[0])

clf.predict(data.data[10])

clf.predict(data.data)

from IPython.display import Image #needed to render in notebook
import StringIO, pydot  #needed to convert dot format to png

dot_data = StringIO.StringIO() 
tree.export_graphviz(clf, out_file=dot_data) 
graph = pydot.graph_from_dot_data(dot_data.getvalue())
Image(graph.create_png())

from sklearn.naive_bayes import GaussianNB

clf2 = GaussianNB()

clf2.fit(data.data, data.target)


from sklearn import cross_validation

kf = cross_validation.KFold(len(data.data), n_folds=10)
len(kf)

pc_folds = []
for train_index, test_index in kf:
    clf2 = GaussianNB()
    clf2.fit(data.data[train_index], data.target[train_index])
    predictions = clf2.predict(data.data[test_index])
    pc_right = sum(predictions == data.target[test_index])/float(len(test_index))
    pc_folds.append(pc_right)
    

pc_folds

mean(pc_folds)

min(pc_folds)

from sklearn import metrics

predictions = clf2.predict(data.data)
predictions

metrics.accuracy_score(data.target, predictions)

print metrics.classification_report(data.target, predictions)

metrics.confusion_matrix(data.target, predictions)

confmat = metrics.confusion_matrix(data.target, predictions)
imshow(confmat,interpolation='nearest', cmap=cm.gray_r)

confmat = metrics.confusion_matrix(data.target, predictions)

imshow(confmat.max() - confmat,interpolation='nearest', cmap=cm.gray)

for rownum, row in enumerate(confmat):
    for colnum, val in enumerate(row):
        text(colnum, rownum, str(val/sum(data.target == rownum).astype(float)), fontsize=24, color='black' if val < confmat.max()/2 else 'white', ha='center', va='center' )
        
xticks(arange(len(data.target_names)), data.target_names)
yticks(arange(len(data.target_names)), data.target_names);