python_机器学习_监督学习模型_决策树

决策树模型练习：https://www.kaggle.com/c/GiveMeSomeCredit/overview

1. 监督学习--分类

机器学习肿分类和预测算法的评估：

　　a. 准确率

　　b.速度

　　c. 强壮行

　　d.可规模性

　　e. 可解释性

2. 什么是决策树/判定树（decision tree）？

https://scikit-learn.org/stable/modules/tree.html

3. 熵（entropy）概念：

变量的不确定越大，熵也就越大。

4. 决策树归纳算法（ID3）

5. 其他算法及优缺点

6. 决策树的应用

生成后的决策树

逻辑代码：

整理好的代码 --》

python3.6.3

Successfully installed joblib-0.13.2 numpy-1.16.4 scikit-learn-0.21.2 scipy-1.3.0

# -*- coding:utf-8 -*-
from sklearn.feature_extraction import DictVectorizer
import csv
from sklearn import preprocessing
from sklearn import tree  # 要求是数值型的值
from sklearn.externals.six import StringIO
import pandas as pd

"""
注意： 决策树要求要数值型的值，不能是字符串类型的值
例如： no， yes这样的值是不允许的
需要转换成矩阵
====================================
age            income    student
youth        high    no
youth        high    no
middle_aged    high    no
senior        medium    no
senior        low    yes
====================================

比如上面这种数据：
youth    middle_aged        senior         high    medium    low    ......
1        0                0            1        0        0
1        0                0            1        0        0
.....

"""

fileName = r"C:\Users\Administrator\Desktop\data.xlsx"
data = pd.read_excel(fileName)
# 删除id序列
del data["RID"]
# headers
headers = data.columns.values
# print(headers)
# ["RID", 'age'.....]

# 样本量
# print(len(data))

# dict格式化单个样本
# print(dict(data.ix[1]))
# 单个样本最后一个数据
# print(data.ix[1][-1])

featureList = []
labelList = []
for row in range(len(data)):
    rowData = data.ix[row]
    labelList.append(rowData[-1])
    featureList.append(dict(rowData))

# print(featureList)

# [
#    {"credit_rating": "fair", "age": "youth"},
#    .... #作用，方便转换成矩阵。将数据转换成对象
# ]
# print(labelList)
# ['no', 'no', 'yes', 'yes', 'yes', 'no', 'yes', 'no', 'yes', 'yes', 'yes', 'yes', 'yes', 'no']

# =========<格式化数据，转换成decision tree需要的格式模型>============

vec = DictVectorizer()
dummyX = vec.fit_transform(featureList).toarray()

print("dummyX:" + str(dummyX))  # 转换成矩阵的数据了二维
print(vec.get_feature_names())

print("labelList: " + str(labelList))

lb = preprocessing.LabelBinarizer()
dummyY = lb.fit_transform(labelList)
print("dummyY: " + str(dummyY))

# ===========《决策树建模分析》=============

clf = tree.DecisionTreeClassifier(criterion="entropy")
clf = clf.fit(dummyX, dummyY)
print("clf: ", str(clf))

# # 存储决策树信息

# # Graphviz 将dot转换成pdf的命令： dot -T pdf iris.dot -o output.pdf
# # 可以查看decision tree 的形状了（看pdf的值）
# with open(r"C:\Users\Administrator\Desktop\code\mechine_learning\allElectronicInformationGainOri.dot", "w") as f:
#     f = tree.export_graphviz(clf, feature_names = vec.get_feature_names(), out_file = f)

# # 下面的代码属于预测的代码
# # 属于转化后的矩阵数值，其实就是进行复制修改
oneRowX = dummyX[2, :]
print("oneRowX: " + str(oneRowX))

newRowX = oneRowX

# newRowX[0] = 1
# newRowX[2] = 1
print("newRowX: ", str(newRowX))

predictedY = clf.predict([newRowX])
# 预测 class_buys_labels的值
print("predictedY: " + str(predictedY))

但这段代码不是特别通用，而且有bug， 需要修改，但基本逻辑是正确的

# -*- coding:utf-8 -*-
from sklearn.feature_extraction import DictVectorizer
import csv
from sklearn import preprocessing
from sklearn import tree  # 要求是数值型的值
from sklearn.externals.six import StringIO

"""
注意： 决策树要求要数值型的值，不能是字符串类型的值
例如： no， yes这样的值是不允许的
需要转换成矩阵
====================================
age            income    student
youth        high    no
youth        high    no
middle_aged    high    no
senior        medium    no
senior        low    yes
====================================

比如上面这种数据：
youth    middle_aged        senior         high    medium    low    ......
1        0                0            1        0        0
1        0                0            1        0        0
.....

"""

allElectronicsData = open(r"C:\Users\Administrator\Desktop\data.xlsx", 'r')
reader = csv.reader(allElectronicsData)
print(reader)
headers = next(reader)

print(headers)
# ["RID", 'age'.....]

featureList = []
labelList = []
for row in reader:
    labelList.append(row[len(row) - 1])
    rowDict = {}
    for i in range(1, len(row) - 1):
        rowDict[headers[i]] = row[i]

    featureList.append(rowDict)

print(featureList)
# [
#    {"credit_rating": "fair", "age": "youth"},
#    .... #作用，方便转换成矩阵。将数据转换成对象
# ]

vec = DictVectorizer()
dummyX = vec.fit_transform(featureList).toarray()

print("dummyX:" + str(dummyX))  # 转换成矩阵的数据了二维
print(vec.get_feature_names())

print("labelList: " + str(labelList))

lb = preprocessing.LabelBinarizer()
dummyY = lb.fit_transform(labelList)
print("dummyY: " + str(dummyY))

clf = tree.DecisionTreeClassifier(criterion="entropy")
clf = clf.fit(dummyX, dummyY)
print("clf: ", str(clf))

# 存储决策树信息

# Graphviz 将dot转换成pdf的命令： dot -T pdf iris.dot -o output.pdf
# 可以查看decision tree 的形状了（看pdf的值）
with open(r"C:\Users\Administrator\Desktop\code\mechine_learning\allElectronicInformationGainOri.dot", "w") as f:
    f = tree.export_graphviz(clf, feature_names = vec.get_feature_names(), out_file = f)

# 下面的代码属于预测的代码
# 属于转化后的矩阵数值，其实就是进行复制修改
oneRowX = dummyX[0, :]
print("oneRowX: " + str(oneRowX))

newRowX = oneRowX

newRowX[0] = 1
newRowX[2] = 0
print("newRowX: ", str(newRowX))

predictedY = clf.predicted(newRowX)
# 预测 class_buys_labels的值
predicted("predictedY: " + str(predictedY))

if __name__ == '__main__':
    main()