def main():
x_train, x_test, y_train, y_test = process_data() # 划分数据集
my_tree = tree_generate(x_train, y_train, max_depth=4,
mini_sample=10) # 构造决策树
precision = evaluate(x_test, y_test, my_tree) # 进行推断和评估
print('准确率 = ', precision)
treePlotter.create_plot(my_tree) # 画出训练的决策树，便于观察结果
def classify_lenses():
with open("lenses.txt", "r") as f:
lenses = [inst.strip().split('\t')
for inst in f.readlines()] # 一行为一个样本（特征、类标签）
lenses_labels = ["age", "prescript", "astigmatic", "testRate"] # 特征标签
lenses_tree = create_tree(lenses, lenses_labels)
print lenses_tree
import treePlotter
treePlotter.create_plot(lenses_tree)
def get_result(self):
"""
得到结果并绘制树
:return:
"""
data_set = self.train_data
labels_tmp = self.attributes[:]
decision_tree = self.create_tree(data_set, labels_tmp)
print("decisionTree:\n", decision_tree)
treePlotter.create_plot(decision_tree)
result = self.classify_all(decision_tree)
print("result:\n", result)
"""
使用pickle存储决策树
:param input_tree: 决策树
:param filename: 需要写入的文件名
:return: None
"""
import pickle
# 以二进制的方式写入文件中
fw = open(filename, 'wb')
pickle.dump(input_tree, fw)
fw.close()
def grab_tree(filename):
"""
读出决策树结构
:param filename: 需要读出的文件名
:return: 返回决策树的内容
"""
import pickle
fr = open(filename, 'rb')
return pickle.load(fr)
if __name__ == '__main__':
fr = open('lenses.txt')
lenses = [inst.strip().split('\t') for inst in fr.readlines()]
lenses_labels = ['age', 'prescript', 'astigmatic', 'tearRate']
lenses_tree = create_tree(lenses, lenses_labels)
treePlotter.create_plot(lenses_tree)
print(lenses_tree)
def test_glass():
fr = open('lenses.txt')
lenses = [line.strip().split('\t') for line in fr.readlines()]
lense_labels = ['age', 'precript', 'astigmatic', 'tearRate']
lense_tree = trees.create_tree(lenses, lense_labels)
treePlotter.create_plot(lense_tree)
def entroy(self, y):
p = pd.value_counts(y) / y.shape[0] # 计算各类样本所占比率
ent = np.sum(-p * np.log2(p))
return ent
if __name__ == '__main__':
data_path2 = 'watermelon2_0_Ch.txt'
data = pd.read_table(data_path2,
encoding='utf8',
delimiter=',',
index_col=0)
train = [1, 2, 3, 6, 7, 10, 14, 15, 16, 17]
train = [i - 1 for i in train]
X = data.iloc[train, :6]
y = data.iloc[train, 6]
test = [4, 5, 8, 9, 11, 12, 13]
test = [i - 1 for i in test]
X_val = data.iloc[test, :6]
y_val = data.iloc[test, 6]
tree = DecisionTree('gini', 'pre_pruning')
tree.fit(X, y, X_val, y_val)
print("平均准确率为:", np.mean(tree.predict(X_val) == y_val))
treePlotter.create_plot(tree.tree_)