Pandas筛选某列过滤的方法

通过dataframe的第二个条件，进行筛选

#make字段异常值清洗
new = data[['make', 'model', 'instance_id']]
new['make_model'] = new['make']+':::'+new['model']
new.head(3)

# new.make_model.value_counts()
# 统计make_model列属性值出现的次数
 
 
 
new.make_model.value_counts()[new.make_model.value_counts() <= 200]
 
"""
OPPO:::OPPO+A59st               200
OPPO:::3007                     200
Xiaomi:::Redmi%20Note%203       200
Meizu:::MEIZU-M6                199
samsung:::SM-N9006              199
                               ... 
OPPO,OPPO A53,A53:::OPPO A53      1
boway Ujs15:::boway U15             1
BaiMao:::BM I8                    1
vivo:::vivoy75a                   1
SUPERJO:::SUPERJO                 1
Name: make_model, Length: 15597, dtype: int64
"""

 找出符合第二列筛选条件的index（这里index不是0-n，而是刚才value_counts()的index）

(new.make_model.value_counts()[new.make_model.value_counts() &编程客栈lt;= 200]).index
 
"""
Index(['OPPO:::OPPO+A59sphpt', 'OPPO:::3007', 'Xiaomi:::Redmi%20Note%203',
       'Meizu:::MEIZU-M6', 'samsung:::SM-N9006', 'Coolpad:::MTS-T0',
       'OPPO R11st:::OPPO R11st', 'Blephone:::lephone T7A', 'GIONEE:::GN9011',
       'Meizu:::PRO 7-S',
       ...
       'HUAWEI:::HUAWEI%25252BG7-UL20', 'VOLTE:::L3', 'GIONEE:::GN868',
       'alps:::SOP-i9', 'GT-I9300I:::GT-I9300I',
       'OPPO,OPPO A53,A53:::OPPO A53', 'boway U15:::boway U15',
       'BaiMao:::BM I8', 'vivo:::vivoy75a', 'SUPERJO:::SUPERJO'],
      dtype='object', length=15597)
"""
 

new.make_model
 
"""
0          HUAWEI:::HUAWEI-CAZ-AL10
1             Xiaomi:::Redmi Note 4
2                  OPPO:::OPPO+R11s
3                               NaN
4                  Apple:::iPhone 7
                     ...           
1041669             OPPO:::OPPO-R9s
1041670              Xiaomi:::MI-5X
1041671             vivo:::vivo Y37
1041672          vivo:::vivo%20Y75A
1041673                  OPPO:::A31
Name: make_model, Length: 1041674, dtype: object
"""

dataframe.loc(行索引, 列名)

# 在make_model列，
# 定位符合 new.make_model.isin((new.make_model.value_counts()[new.make_model.value_counts() <= 200]).index) 的行
 
# 
 
new.loc[new.make_model.isin((new.make_model.value_counts()[new.make_model.value_counts() <= 200]).index), 'make_model'] = 'other' #去除低频词

 再感受下第二个case

data['day'] = data['time'].apply(lambda x : int(time.strftime("%d", time.localtime(x))))
data['period'] = data['day']
data[['period']].head(3)

data['period'].unique()
 
# array([29, 30, 31, 27,  1,  2, 28,  3])

 直接用列筛选

[dawww.devze.comta['period']<27]
 
"""
[0          False
 1          False
 2          False
 3          False
 4          False
            ...  
 1041669     True
 1041670     True
 1041671     True
 1041672     True
 1041673     True
 Name: period, Length: 1041674, dtype: bool]
"""

 
data['period']<27
 
"""
0开发者_开发培训          False
1          False
2          False
3          False
4          False
           ...  
1041669     True
1041670     True
1041671     True
1041672     True
1041673     True
Name: period, Length: 1041674, dtype: bool
"""

挑选period列，值<27的行（已成功挑选）

data['period'][data['period']<27]
 
"""
950        1
951        1
952        1
953        1
954        1
          ..
1041669    3
1041670    3
1041671    3
1041672    3
1041673    3
Name: period, Length: 348536, dtype: int64
"""
 
 
data['period'][data['period']<27] = data['period'][data['period']<27] + 31

这样可以使用head展示

data[['period']][data['period']<27].head(3)

还有种单列就能筛选的方法

t2['receive_number'] = t2.date_received.apply(lambda s:len(s.split(':')))
t2 = t2[t2.receive_number>1]
t2.head(3)

到此这篇关于Pandas筛选某列过滤的方法的文章就介编程客栈绍到这了,更多相关Pandas筛选某列过滤内容请搜索我们以前的文章或继续浏览下面的相关文章希望大家以后多多支持我们！