1 %pylab inline
2 %matplotlib inline
3 import seaborn as sns
4 import numpy as np
5 import pandas as pd
6 import matplotlib.pyplot as plt
7 titanic_data = pd.read_csv('titanic-data.csv')
8 titanic_data.info()

1 #统计变量和生存率的关系,如果需要使用堆栈图更清晰的展示数据,stacked的值设置为True,为False默认展示该变量下的生存率
2 def visualize_survival(feature,stacked=False):
3     if stacked:
4         survived_rate = titanic_data.groupby([feature,'Survived'])['Survived'].count().unstack().plot(kind='bar',stacked='True')
5     else:
6         survived_rate = (titanic_data.groupby([feature]).sum()/titanic_data.groupby([feature]).count())['Survived']
7         survived_rate.plot(kind='bar')
8     plt.title(feature + ' V.S. Survival') 

1 #比较单个变量之间的关系,feature表示要分析的列,args表示x轴的名称
2 def visualize_column(feature,*args):
3     fig,ax=plt.subplots(figsize=(7,5))
4     titanic_data[feature].value_counts().plot(kind='bar')
5     for i in range(len(args)):
6         ax.set_xticklabels((args[i]),rotation='horizontal')
7     ax.set_title('bar of ' + feature)

1 fig,axes=plt.subplots(2,1,figsize=(15,5))
2 titanic_data['Age'].hist(ax=axes[0])  #年龄分布
3 axes[0].set_title('plot of age')
4 titanic_data['Fare'].hist(ax=axes[1])  #票价分布
5 axes[1].set_title('plot of fare')

1 survived_rate = float(titanic_data['Survived'].sum())/titanic_data['Survived'].count()
2 print survived_rate
3 by_survived = titanic_data.groupby(['Survived'])['Survived'].count()
4 plt.pie(by_survived,labels=['Non-Survived','Survived'],autopct='%1.0f%%')
5 plt.title('Pie Chart Of Surviveness for Surviveness of Passengers')

1 visualize_column('Sex',('Male','Female'))

1 visualize_column('Pclass',('Class 3','Class 2','Class 1')) 

1 visualize_survival('Sex',True)

1 visualize_survival('Pclass',True)

1 titanic_data.Age.fillna(titanic_data.Age.mean(),inplace=True)   #使用均值来填充Age中的空值
2 ages = np.arange(0,90,10) #年龄分段
3 titanic_data['age_cut'] = pd.cut(titanic_data.Age,ages)
4 visualize_survival('age_cut',True)

1 titanic_data.Embarked[titanic_data.Embarked.isnull()]
2 print titanic_data[titanic_data['Embarked'].isin(['S','C','Q'])==False]
3 titanic1 = titanic_data[titanic_data['Pclass']==1]
4 ax=sns.boxplot(titanic1.Embarked,titanic1.Fare)
5 plt.plot((-100,100),(80,80),'r-')

1 titanic_data.Embarked=titanic_data.Embarked.fillna('C')
2 visualize_survival('Embarked')

1 fares = np.arange(0,600,50) #划分票价区间
2 fares_cut = pd.cut(titanic_data.Fare,fares)
3 fares_cut_group = titanic_data.groupby(fares_cut)
4 fares_cut_group.count().PassengerId #获取异常数据
5 titanic_data.Fare.sort_values(ascending=False).head() #查看异常数据
6 #进行IQR运算找出异常数据
7 q75,q25 = np.percentile(titanic_data.Fare,[75,25])
8 iqr = q75-q25
9 print q75+iqr*3  #确定异常数据的值

1 fares_count = titanic_data.Fare[titanic_data.Fare<100.27] #舍弃异常数据
2 fares_count_range = np.arange(0,110,10) #重新计算票价区间
3 titanic_data['new_fare'] = pd.cut(titanic_data.Fare,fares_count_range)
4 visualize_survival('new_fare')

1 #通过SibSp+Parch总体计算出陪伴的生还率
2 titanic_data['family_member'] = titanic_data.SibSp+titanic_data.Parch
3 visualize_survival('family_member')