【sklearn】数据预处理 sklearn.preprocessing

data = [[-1, 2], [-0.5, 6], [0, 10], [1, 18]]

1. 归一化 sklearn.preprocessing.MinMaxScaler

from sklearn.preprocessing import MinMaxScaler
# 归一化 默认范围[0,1]
scaler = MinMaxScaler(feature_range=[5, 10])
# 数据特征过多.fit()报错，用.partial_fit()
scaler = scaler.fit(data)
result = scaler.transform(data)
# 也可训练导出结果一步达成
# result = scaler.fit_transform(data)
result
# 逆转归一化回原数据
# scaler.inverse_transform(result)

'''
    array([[ 5.  ,  5.  ],
           [ 6.25,  6.25],
           [ 7.5 ,  7.5 ],
           [10.  , 10.  ]])
'''

2. 标准化 sklearn.preprocessing.StandardScaler

from sklearn.preprocessing import StandardScaler
# 标准化
scaler = StandardScaler()
scaler.fit(data)
print('两列均值：',scaler.mean_)
print('两列方差：',scaler.var_)
x_std = scaler.transform(data)
print('均值:',x_std.mean(),'方差:',x_std.std())
print(x_std)

'''
    两列均值： [-0.125  9.   ]
    两列方差： [ 0.546875 35.      ]
    均值: 0.0 方差: 1.0
    [[-1.18321596 -1.18321596]
     [-0.50709255 -0.50709255]
     [ 0.16903085  0.16903085]
     [ 1.52127766  1.52127766]]
 '''

3. 缺失值 sklearn.impute.SimpleImputer

查看数据

import pandas as pd
# index_col=0 第0列是索引
data = pd.read_csv('./data_preprocessing.csv', index_col=0)

data.info()
'''
    <class 'pandas.core.frame.DataFrame'>
    Int64Index: 9 entries, 0 to 8
    Data columns (total 4 columns):
     #   Column    Non-Null Count  Dtype  
    ---  ------    --------------  -----  
     0   Age       7 non-null      float64
     1   Sex       9 non-null      object 
     2   Embarked  9 non-null      object 
     3   Survived  9 non-null      object 
    dtypes: float64(1), object(3)
    memory usage: 360.0+ bytes
'''

填补缺失值

# 填补缺失值
Age = data.loc[:,'Age'].values
# 不能是一维
Age = Age.reshape(-1, 1)

from sklearn.impute import SimpleImputer
# strategy 默认均值填补
imp_median = SimpleImputer(strategy='median')
imp_0 = SimpleImputer(strategy='constant', fill_value=0)

imp_median = imp_median.fit_transform(Age)
imp_0 = imp_0.fit_transform(Age)
print('中位数填补:',imp_median,
     '\n',
      '0填补:',imp_0
     )
'''
    中位数填补: [[22.]
     [38.]
     [26.]
     [26.]
     [35.]
     [26.]
     [58.]
     [20.]
     [ 2.]] 
     0填补: [[22.]
     [38.]
     [26.]
     [ 0.]
     [35.]
     [ 0.]
     [58.]
     [20.]
     [ 2.]]
'''

# 众数填补
Embarked = data.loc[:,'Embarked'].values.reshape(-1, 1)
imp_mode = SimpleImputer(strategy='most_frequent')
data.loc[:,'Embarked'] = imp_mode.fit_transform(Embarked)

pandas numpy简单的处理方法

# pandas numpy处理
# .fillna() Dataframe中直接填补
data.loc[:,'Age'].fillna(data.loc[:,'Age'].mean())
# axis=1 删除缺失值所在列
data.dropna(axis=0, inplace=False)

	Age	Sex	Embarked	Survived
0	22.0	male	S	No
1	38.0	female	C	Yes
2	26.0	female	S	Yes
4	35.0	male	S	No
6	58.0	male	Q	No
7	20.0	female	C	Yes
8	2.0	female	S	Unknown