# 使用SimpleImputer

from sklearn.impute import SimpleImputer

# 使用中位数填补Age

num = ['Age']

data_num = data.loc[:, num]

SI_num = SimpleImputer(strategy = 'median')

SI_num = SI_num.fit(data_num)

data.loc[:, num] = SI_num.transform(data_num)

# 使用众数填补Embarked

cat = ['Sex', 'Embarked']

data_cat = data.loc[:, cat]

SI_cat = SimpleImputer(strategy = 'most_frequent')

SI_cat = SI_cat.fit(data_cat)

data.loc[:, cat] = SI_cat.transform(data_cat)

# 对Y值进行空缺值填补

lab = ['Survived']

data_lab = data.loc[:, lab]

SI_lab = SimpleImputer(strategy = 'most_frequent',

missing_values = 'Unknown')

SI_lab = SI_lab.fit(data_lab)

data.loc[:, lab] = SI_lab.transform(data_lab)