Pandas Dataframe 中多列的多个 Elseif 条件答案

【问题标题】：multiple Elseif conditions in Pandas Dataframe over multiple columnsPandas Dataframe 中多列的多个 Elseif 条件
【发布时间】：2021-03-18 16:27:21
【问题描述】：

我在对大型数据集应用多个 Elif 条件时遇到问题。样本数据如下：

Id = ['AM12','AM21','AM31','AM41','AM66','AM81','AM77','AM87','AM27','AM69']
Exec = ['Athreyu','Megan','','Omar','Michael','','Oliver','','Jesus','']
AD_Executive= ['','Fer','Virat','John','','John','','','pandya','John']
Ex_FiscId= ['John','Sonal','','Ram','','Anthony','','','Sriju','']
full_nm = ['pulari','','Burgers','Saheb','Bhavya','Borah','Dutta','Upinder','Ruhaan','Rochan']
df_ex = pd.DataFrame(list(zip(Id, full_nm,Exec,AD_Executive,Ex_FiscId)), 
           columns =['Id', 'full_nm','Exec','AD_Executive','Ex_FiscId'])

我想为最终名称创建一个新列。我申请的条件是 -

def final(df_ex):
   if df_ex['Ex_FiscId'] != np.NaN:
    return df_ex['Ex_FiscId']
  elif (df_ex['Ex_FiscId'] == np.NaN) & (df_ex['AD_Executive'] != np.NaN):
    return df_ex['AD_Executive']
  elif (df_ex['Ex_FiscId'] == np.NaN) & (df_ex['AD_Executive'] == np.NaN) & (df_ex['Exec'] !=np.NaN):
    return df['Exec']
  elif (df_ex['Ex_FiscId'] == np.NaN) & (df_ex['AD_Executive'] == np.NaN) & (df_ex['Exec'] ==np.NaN):
    return df_ex['full_nm']

df_ex['Final'] = df_ex.apply(final, axis = 1)

但它没有产生所需的输出。该代码似乎只读取了第一个 if 条件而忽略了其他条件。

我还附上了输入和所需输出表以供参考-

Id	full_nm	Exec	AD_Executive	Ex_FiscId
AM12	pulari	Athreyu		John
AM21		Megan	Fer	Sonal
AM31	Burgers		Virat
AM41	Saheb	Omar	John	Ram
AM66	Bhavya	Michael
AM81	Borah		John	Anthony
AM77	Dutta	Oliver
AM87	Upinder
AM27	Ruhaan	Jesus	pandya	Sriju
AM69	Rochan		John

期望的输出 -

Id	full_nm	Exec	AD_Executive	Ex_FiscId	Final
AM12	pulari	Athreyu		John	John
AM21		Megan	Fer	Sonal	Sonal
AM31	Burgers		Virat		Virat
AM41	Saheb	Omar	John	Ram	Ram
AM66	Bhavya	Michael			Michael
AM81	Borah		John	Anthony	Anthony
AM77	Dutta	Oliver			Oliver
AM87	Upinder				Upinder
AM27	Ruhaan	Jesus	pandya	Sriju	Sriju
AM69	Rochan		John		John

【问题讨论】：

所有内容，包括np.NaN，都是!= np.NaN，所以你永远不会超过第一个块。您应该使用pd.isna() 检查某些内容是否为空。但实际上你应该使用np.select 实现所有这些逻辑：参见stackoverflow.com/questions/18194404/…

标签： pandas data-manipulation

【解决方案1】：

让我们试试这个：

df['final'] = df_ex.apply(lambda x: x[x[::-1].notnull().idxmax()], axis=1)

输出：

    Id  full_nm     Exec    AD_Executive    Ex_FiscId   final
0   AM12    pulari  Athreyu     NaN            John     John
1   AM21    NaN     Megan       Fer            Sonal    Sonal
2   AM31    Burgers NaN         Virat          NaN      Virat
3   AM41    Saheb   Omar        John           Ram      Ram
4   AM66    Bhavya  Michael     NaN            NaN      Michael
5   AM81    Borah   NaN         John           Anthony  Anthony
6   AM77    Dutta   Oliver      NaN            NaN      Oliver
7   AM87    Upinder NaN         NaN            NaN      Upinder
8   AM27    Ruhaan  Jesus       pandya         Sriju    Sriju
9   AM69    Rochan  NaN         John           NaN      John

【讨论】：

【解决方案2】：

def final(df_ex):
if df_ex['Ex_FiscId'] !="":
    return df_ex['Ex_FiscId']
elif (df_ex['Ex_FiscId']=="") & (df_ex['AD_Executive'] !=""):
    return df_ex['AD_Executive']
elif (df_ex['Ex_FiscId']=="") & (df_ex['AD_Executive'] =="") & (df_ex['Exec'] !=""):
    return df_ex['Exec']
elif (df_ex['Ex_FiscId']=="") & (df_ex['AD_Executive'] =="") & (df_ex['Exec'] ==""):
    return df_ex['full_nm']


df_ex['Final'] = df_ex.apply(final, axis = 1)

【讨论】：