Python数据框-来自3列的数据值将映射到其他3列下

Question

Python experts, I need some help understanding how I can re-shape or pivot columns of my pandas dataframe. Following is what I want to do, I have a situation where I have 6 columns First_name1, middle_name1, last_name1, first_name2, middle_name2 and last_name2. I want to bring the values of the last 3 columns under the 1st 3 columns, like some sort of pivoting but not exactly.

First Name 1    Middle Name 1   Last Name 1 First Name 2    Middle Name 2   Last Name 2
    ABC      DAF                 XYZ          LOU           NaN      XYZ

I am unable to understand how to do it. I only need 1st 3 columns existing after this operation. Any help would be certainly appreciated.

Answer 1

Use rsplit for MultiIndex and then reshape by stack , last reset_index :

df.columns = df.columns.str.rsplit(n=1, expand=True)
print (df)
  First Name Middle Name Last Name First Name Middle Name Last Name
           1           1         1          2           2         2
0        ABC         DAF       XYZ        LOU         NaN       XYZ

df = df.stack().reset_index(drop=True)
print (df)
  First Name Last Name Middle Name
0        ABC       XYZ         DAF
1        LOU       XYZ         NaN

Another solution with undocumented function lreshape :

d = {'First Name':df.columns[df.columns.str.startswith('First Name')],
     'Middle Name':df.columns[df.columns.str.startswith('Middle Name')],
     'Last Name':df.columns[df.columns.str.startswith('Last Name')]}

df = pd.lreshape(df, d, dropna=False)
print (df)
  Middle Name First Name Last Name
0         DAF        ABC       XYZ
1         NaN        LOU       XYZ

Answer 2

You can use the dataframe construtor and values reshape ie

ndf = pd.DataFrame(df.values.reshape(-1,3),columns=['First Name','Middle Name','Last Name'])

First Name Middle Name Last Name
0        ABC         DAF       XYZ
1        LOU         NaN       XYZ

Sample run on multiple row data ie

First Name 1 Middle Name 1 Last Name 1 First Name 2  Middle Name 2 Last Name 2
0        ABC         DAF       XYZ        LOU          NaN       XYZ
0        ABC         DAF       XYZ        LOU          NaN       XYZ

Output :

First Name Middle Name Last Name
0        ABC         DAF       XYZ
1        LOU         NaN       XYZ
2        ABC         DAF       XYZ
3        LOU         NaN       XYZ

Answer 3

This is a typically pd.wide_to_long question , one line code as below :

pd.wide_to_long(df.reset_index(), ["FirstName", "MiddleName","LastName"], i="index", j="year").reset_index(drop=True)
Out[506]: 
  FirstName MiddleName LastName
0       ABC        DAF      XYZ
1       ABC        DAF      XYZ
2       LOU        NaN      XYZ
3       LOU        NaN      XYZ