python 泰坦尼克存活问题分析

先导入基本的模块

#导入模块
import numpy as np
import pandas as pd
from scipy import stats
import seaborn as sns
import matplotlib.pyplot as plt
%matplotlib inline
plt.rc("font",family="SimHei",size="12")  #解决中文无法显示的问题

导入数据

#导入CSV数据
train=pd.read_csv('F://python//titanic_data.csv')

字段含义：

PassengerId：乘客编号

Survived：乘客是否存活

Pclass：乘客所在的船舱等级

Name：乘客姓名

Sex：乘客性别

Age：乘客年龄

SibSp：乘客的兄弟姐妹和配偶数量

Parch：乘客的父母与子女数量

Ticket：票的编号

Fare：票价

Cabin：座位号

Embarked：乘客登船码头

查查数据分布

#使用info，describe 
train.info()
train.describe()

查看缺失值

#null
train.isnull().sum()

区别类别变量和连续变量

#连续变量和类别变量分开
num_features=train.select_dtypes(include=[np.number])
categ_features=train.select_dtypes(include=[np.object])
num_features=num_features.columns
categ_features=categ_features.columns

类别变量分析（['Name', 'Sex', 'Ticket', 'Cabin', 'Embarked']）主要分析性别和乘客登船码头

#类别变量的每个类别频数可视化
def count_plot(x,  **kwargs):
    sns.countplot(x=x)
    x=plt.xticks(rotation=90)

f = pd.melt(train,  value_vars=['Sex','Embarked'])
g = sns.FacetGrid(f, col="variable",  col_wrap=2, sharex=False, sharey=False, size=5)
g = g.map(count_plot, "value")

查看这二者和存活的关系如何

#画图
pd.crosstab(train["Sex"],train["Survived"]).plot(kind="bar")
pd.crosstab(train['Embarked'],train["Survived"]).plot(kind="bar")

处理连续变量（['PassengerId', 'Survived', 'Pclass', 'Age', 'SibSp', 'Parch', 'Fare']）

#每个数字特征得分布可视化
f = pd.melt(train, value_vars=num_features)
g = sns.FacetGrid(f, col="variable",  col_wrap=2, sharex=False, sharey=False)
g = g.map(sns.distplot, "value")

存活画图

#使用pandas的画图
train["Survived"].value_counts().plot(kind="bar")

乘客所在的船舱等级

#乘客所在的船舱等级画图
train['Pclass'].value_counts().plot(kind="bar")

#和存活做二维表画图
pd.crosstab(train['Pclass'],train["Survived"]).plot(kind="bar")

 年龄处理

#画条形图
train["Age"].hist()

#按照五岁一个区间去分
cut=[i*5 for i in range(1,20)]
train.age=pd.cut(train.Age,cut)

a=pd.crosstab(train.age,train.Survived)
a.plot(kind="bar")
a['rate']=a[1]/(a[1]+a[0])
a['rate'].plot(kind="bar")

效果并不明显，再分一次

#重新分区间
cut=[0,5,15,20,30,40,50,60,100]
train.age=pd.cut(train.Age,cut)
a=pd.crosstab(train.age,train.Survived)
a.plot(kind="bar")
a['rate']=a[1]/(a[1]+a[0])
a['rate'].plot(kind="bar")

 

SibSp：乘客的兄弟姐妹和配偶数量

Parch：乘客的父母与子女数量

 pd.crosstab(train["SibSp"],train["Survived"]).plot(kind="bar")
pd.crosstab(train["Parch"],train["Survived"]).plot(kind="bar")

 票价

#使用箱型图查看二者分布如何
fig,ay = plt.subplots()
Fare1 = train.Fare[train.Survived == 1]
Fare0 = train.Fare[train.Survived == 0]
plt.boxplot((Fare1,Fare0),labels=('Survived','Not Survived'))
ay.set_ylim([-10,150])
ay.set_title("Boxplot of Fare")

train["Fare"].hist()

变量之间相关性分析

#corr
corr=train.corr()
corr["Survived"].sort_values()

特征构造

#缺失值处理
train['Age'].mean()  #29.69911764705882
train['Age'].fillna(30,inplace=True)

#删除
train['Cabin'].value_counts()
train=train.drop(['Cabin'],axis=1)
#众数填充
train['Embarked'].value_counts()
train['Embarked'].fillna('S',inplace=True)


#年龄分箱
train.age=pd.cut(train.Age,[0,5,15,20,35,50,60,100])
pd.crosstab(train.age,train.Survived).plot.bar()

train.Parch[(train.Parch>0) & (train.Parch<=2)]=1
train.Parch[train.Parch>2]=2
train.SibSp[(train.SibSp>0) & (train.SibSp<=2)]=1
train.SibSp[train.SibSp>2]=2



dummy_Pclass = pd.get_dummies(train.Pclass, prefix='Pclass')
dummy_Sex = pd.get_dummies(train.Sex, prefix='Sex')
dummy_Embarked = pd.get_dummies(train.Embarked, prefix='Embarked')
dummy_Parch = pd.get_dummies(train.Parch, prefix='Parch')
dummy_SibSp = pd.get_dummies(train.SibSp, prefix='SibSp')
dummy_Age = pd.get_dummies(train.age, prefix='Age')


train_1=dummy_Pclass.join(dummy_Sex).join(dummy_Embarked).join(dummy_Parch).join(dummy_SibSp).join(dummy_Age)
train_1['Fare']=train['Fare']

模型构造

from sklearn.linear_model import LogisticRegression
from sklearn.metrics import confusion_matrix, roc_curve,roc_auc_score,classification_report 
from sklearn.model_selection import train_test_split

train_x=train_1
train_y=train['Survived']
x_train,x_test,y_train,y_test=train_test_split(train_x,train_y,test_size=0.3,random_state=0)


#逻辑回归
clf = LogisticRegression()
clf.fit(x_train,y_train)

#用测试集进行检验
clf.predict(x_test)

#混淆矩阵
confusion_matrix(y_test,clf.predict(x_test))  
#array([[144,  24],[ 28,  72]], dtype=int64)

#roc
roc_auc_score(y_test,clf.predict_proba(x_test)[:,1])   #0.8578869047619048

#画图
fpr,tpr,thresholds = roc_curve(y_test,clf.predict_proba(x_test)[:,1])
plt.plot(fpr,tpr)


#分类报告
print(classification_report(y_test,clf.predict(x_test)))