decision_tree_cost_complexity

from sklearn import tree
from sklearn.model_selection import train_test_split
import numpy as np
import ipdb


np.random.seed(0)

X = np.random.random([500,2])
Y = []
for x_value in X:
    y_value = 0
    if (x_value[0] > 0.5) and (x_value[1] > 0.5):  # AND "logic"
        y_value = 1

    #random part
    #if np.random.random() > 0.9:
    if np.random.random() > 0.9:
        y_value = not y_value
    Y.append(y_value)

Y = np.array(Y)

feature_names = ['x1','x2']
target_names = ['falso','verdadeiro']


X_train, X_test, Y_train, Y_test = train_test_split(X, Y, test_size = 0.2, random_state = 0)
# criar a árvore inicial
clf = tree.DecisionTreeClassifier()

# encontrar os elos fracos (valores de alfa onde as "mudanças ocorrem")
path = clf.cost_complexity_pruning_path(X_train, Y_train)
ccp_alphas, impurities = path.ccp_alphas, path.impurities

# criar uma árvore para cada valor de alfa
clfs = []
for ccp_alpha in ccp_alphas:
    clf = tree.DecisionTreeClassifier(random_state=0, ccp_alpha=ccp_alpha)
    clf.fit(X_train, Y_train)
    clfs.append(clf)

train_scores = [clf.score(X_train, Y_train) for clf in clfs]
test_scores = [clf.score(X_test, Y_test) for clf in clfs]

ipdb.set_trace()


# escolher a melhor


#.
#.
#.

clf = clf.fit(X_train, Y_train)

# export in graphical format
import graphviz
dot_data = tree.export_graphviz(clf, out_file=None, filled=False, rounded=True, impurity=True,
class_names=target_names,
                                feature_names=feature_names
)

graph = graphviz.Source(dot_data)
graph.render("graph")

# export in text format
r = tree.export_text(clf)
print('\n'+r)