基于决策树的泰坦尼克号幸存者分析

import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
%matplotlib inline
import seaborn as sns
sns.set() # setting seaborn default for plots

from sklearn.tree import DecisionTreeClassifier    # 分类决策树
from sklearn.model_selection import GridSearchCV   # 网格搜索
from sklearn.model_selection import train_test_split # 训练集和测试集的划分
from sklearn.model_selection import cross_val_score  # 交叉验证模块

data = pd.read_csv("/Users/peter/data-visualization/train.csv")

# 去掉缺失值过多的列 Cabin （直接删掉），以及和观测判断来说和预测的y没有关系的列（Name, Ticket）
data.drop(["Cabin", "Name", "Ticket"], inplace=True, axis=1)

# Age属性用中值填充
data['Age'] = data['Age'].fillna(data['Age'].mean())

data = data.dropna()   # 只要是存在的缺失值都给删掉
data.info()

# 方法1
# 属性的结果存在多个不同的取值
labels = data["Embarked"].unique().tolist()  # ["S", "C", "Q"]
# 将Embarked属性中的 ["S", "C", "Q"]转成0，1，2
data['Embarked']= data['Embarked'].apply(lambda x: labels.index(x))


# 方法2
data["Sex"] = (data["Sex"] == "male").astype("int")  # 先获取bool值，再将bool值转成0-1

# 方法3：通过loc来实现
# data.loc[:, "Sex"] = (data["Sex"] == "male").astype("int")

x = data.iloc[:, data.columns != "Survived"]
y = data.iloc[:, data.columns == "Survived"]

# 划分训练集和测试集
Xtrain, Xtest, ytrain, ytest = train_test_split(x, y, test_size=0.3)

for i in [Xtrain, Xtest, ytrain, ytest]:
    i.index = range(i.shape[0])  # shape属性的第一个元素就是索引的总个数

clf = DecisionTreeClassifier(random_state=25)
clf = clf.fit(Xtrain, ytrain)
score = clf.score(Xtest, ytest)
score

# 交叉验证
clf = DecisionTreeClassifier(random_state=25)
score = cross_val_score(clf, x, y, cv=10).mean()  # 10次交叉验证求均值，结果降低
score

tr = []  # 两个空列表用来存储训练得到的两个分数
te = []
for i in range(10):
    clf = DecisionTreeClassifier(random_state=25   # 实例化模型，设定决策树的最大深度
                                ,max_depth=i+1
                                ,criterion="entropy"  # 默认是基尼系数。一般情况下，entropy是欠拟合的情况下才使用
                                )
    clf = clf.fit(Xtrain, ytrain)  # fit训练过程
    score_tr = clf.score(Xtrain, ytrain)  # 训练集score
    score_te = cross_val_score(clf, x, y, cv=10).mean()   # 测试集分数，取均值
    tr.append(score_tr)
    te.append(score_te)

print(max(te))
plt.plot(range(1,11), tr, color="red", label="train")  # x,y,颜色，label
plt.plot(range(1,11), te, color="blue", label="test")
plt.xticks(range(1,11)) # 横坐标范围固定在1-10
plt.legend()  # 图例
plt.show()

gini_threholds = np.linspace(0, 0.5, 20)   # gini:0-0.5
#entropy_threholds = np.linspace(0,1,20)  entropy:0-1

# 参数
parameter = {
    "criterion":("gini", "entropy")
    ,"splitter":("best", "random")
    ,"max_depth":[*range(1,10)]
    ,"min_samples_leaf":[*range(1,50,5)]
    ,"min_impurity_decrease":gini_threholds
}

# 网格搜索
clf = DecisionTreeClassifier(random_state=25)
GS = GridSearchCV(clf, parameter, cv=10)  # 同时满足fit,score,交叉验证3种功能
GS = GS.fit(Xtrain, ytrain)

# 2个重要的属性
GS.best_params_   # 显示属性的最好组合结果
GS.best_score_