df和series和rename

介绍下本人。中山大学，医学生+计科学生的集合体，机器学习爱好者。
一、series和df区别

s1=pd.Series([1,2,4,6,7,2])
print(s1)
# 0    1
# 1    2
# 2    4
# 3    6
# 4    7
# 5    2
# dtype: int64

s2=pd.Series([4,3,1,57,8],index=['a','b','c','d','e'])
print(s2)
# a     4
# b     3
# c     1
# d    57
# e     8
# dtype: int64

s3=pd.DataFrame([1,2,4,6,7,2])
print(s3)
#    0
# 0  1
# 1  2
# 2  4
# 3  6
# 4  7
# 5  2

s4=pd.DataFrame([4,3,1,57,8],index=['a','b','c','d','e'],columns=['A'])
print(s4)
#     A
# a   4
# b   3
# c   1
# d  57
# e   8

二、series和df的rename

#series和df在rename的时候，加columns=，然后冒号后面是修改后的名字
train_nuni1 = train.nunique().reset_index()
print(train_nuni1.head())
#                      index            0
# 0             instance_id       1001650
# 1                    time        443670
# 2                    city           333
# 3                province            35
# 4               user_tags        464200
# 5                 carrier             4

train_nuni2 = train.nunique().reset_index().rename(columns={'index':'feat',0: 'data_nunique'})
print(train_nuni2.head())
#                      feat  data_nunique
# 0             instance_id       1001650
# 1                    time        443670
# 2                    city           333
# 3                province            35
# 4               user_tags        464200
# 5                 carrier             4

adid_cnt1=data.groupby(['instance_id'])['adid'].count().reset_index()
print(adid_cnt1.head())
#                  instance_id  adid
# 0             12204730147108     1
# 1             33640746791159     1
# 2             37672601301929     1
# 3             63921594271660     1
# 4             65319037300634     1

adid_cnt2=data.groupby(['instance_id'])['adid'].count().reset_index().rename(columns={'instance_id':'instance_id1','adid':'adid_cnt'})
print(adid_cnt2.head())
#      instance_id1  adid_cnt
# 0  12204730147108         1
# 1  33640746791159         1
# 2  37672601301929         1
# 3  63921594271660         1
# 4  65319037300634         1

adid_cnt3=data.groupby(['instance_id'])['adid'].count().reset_index().rename({'instance_id':'instance_id1','adid':'adid_cnt'})
print(adid_cnt3.head())
#       instance_id  adid
# 0  12204730147108     1
# 1  33640746791159     1
# 2  37672601301929     1
# 3  63921594271660     1
# 4  65319037300634     1

adid_cnt4=data.groupby(['instance_id'])['adid'].count().reset_index(drop_index=True).rename({'instance_id':'instance_id1','adid':'adid_cnt'})
print(adid_cnt4.head())
#TypeError: reset_index() got an unexpected keyword argument 'drop_index'

三、注意type

train_nuni1=train.nunique()#series
print(type(train_nuni1))#<class 'pandas.core.series.Series'>

train_nuni2 = train.nunique().reset_index()#dataframe
print(type(train_nuni2))#<class 'pandas.core.frame.DataFrame'>

四、df和series的合并

train_nuni = train.nunique().reset_index().rename(columns={'index':'feat',0: 'data_nunique'})
print(train_nuni.head())
#           feat  data_nunique
# 0  instance_id       1001650
# 1         time        443670
# 2         city           333
# 3     province            35
# 4    user_tags        464200

test_nuni = test.nunique().reset_index().rename(columns={'index':'feat',0: 'data_nunique'})
print(test_nuni.head())
#           feat  data_nunique
# 0  instance_id         40024
# 1         time         30639
# 2         city           332
# 3     province            35
# 4    user_tags         20237

data_nuni = data.nunique().reset_index().rename(columns={'index':'feat1',0: 'data_nunique1'})
print(data_nuni.head())
#                    feat1  data_nunique1
# 0                   adid           2113
# 1              advert_id             39
# 2  advert_industry_inner             25
# 3            advert_name             35
# 4            app_cate_id             23

df_nunique=pd.merge(train_nuni,test_nuni,'left').rename(columns={'feat':'feat1','data_nunique': 'data_nunique1'})
print(df_nunique.head())
#                     feat1  data_nunique1
# 0             instance_id        1001650
# 1                    time         443670
# 2                    city            333
# 3                province             35
# 4               user_tags         464200

df_nunique=pd.merge(df_nunique,data_nuni,'left').rename(columns={'feat1':'feat2','data_nunique1': 'data_nunique2'})
print(df_nunique.head())
#                     feat2  data_nunique2
# 0             instance_id        1001650
# 1                    time         443670
# 2                    city            333
# 3                province             35
# 4               user_tags         464200

df_nuni=pd.DataFrame({'train':train.nunique(),'test':test.nunique(),'data':data.nunique()})
print(df_nuni)
#                           data     test    train
# adid                      2113    881.0     2079
# advert_id                   39     31.0       38
# advert_industry_inner       25     21.0       24
# advert_name                 35     29.0       34
# app_cate_id                 23     20.0       22
# app_id                     440    174.0      438
# app_paid                     1      1.0        1
# campaign_id                 66     45.0       64

df_nuni=pd.DataFrame({'train':train.nunique(),'test':test.nunique(),'data':data.nunique()}).reset_index()
print(df_nuni)
#                     index     data     test    train
# 0                    adid     2113    881.0     2079
# 1               advert_id       39     31.0       38
# 2   advert_industry_inner       25     21.0       24
# 3             advert_name       35     29.0       34
# 4             app_cate_id       23     20.0       22
# 5                  app_id      440    174.0      438

总结至此，不足之处，批评指正~