CODEWITHSUNDEEP
pythondictionarypandastuples
calpyte
calpyte

Reputation: 885

Pandas DataFrame column from a tuple

I have a dictionary:

employer =  
{'CrntEmp_city': ('XXX', 'XXX'),
'CrntEmp_cntry': ('XXX', 'XXX'),
'CrntEmp_orgNm': ('XXXX LLC', 'YYYY LLC'),
'CrntEmp_orgPK': ('1234567891', '1234567899'),
'CrntEmp_postlCd': ('12345', '12345'),
'CrntEmp_state': ('AK', 'AK'),
'CrntEmp_str1': ('999 XXX', '999 XXX'),
'CrntEmp_str2': ('XXXX', 'XXXX')}

and I want to read this into a DataFrame with one row and the eight columns corresponding to the keys. But however I try this, pandas will always give me two columns (by splitting the tuple where the comma is). Example:

pd.DataFrame([tuple(i) for i in employ.values()])


returns 
    0           1
0   XXXX LLC    YYYY LLC
1   XXXX        XXXX
2   999 XXX     999 XXX
3   XXX         XXX
4   12345       12345
5   XXX         XXX
6   AK          AK
7   1234567891  1234567899

What I want is this:

     0                    1
 0  (XXXX LLC, YYYY LLC)  (XXXX, XXXX)
 etc.

Hence this command fails:

pd.DataFrame([tuple(i) for i in  employ.values()],columns=employ.keys(),index=[0])

Any ideas?

Upvotes: 2

Views: 1125

Answers (3)

Alexander
Alexander

Reputation: 109546

You basically want something like: pd.DataFrame({'col1': [(a, b)], 'col2': [(c, d)]})

You can achieve that using a dictionary comprehension as follows in Python 2. For Python 3, use employer.iter() instead.

>>> pd.DataFrame({k: [tuple(v)] for k, v in employer.iteritems()})

  CrntEmp_city CrntEmp_cntry         CrntEmp_orgNm             CrntEmp_orgPK CrntEmp_postlCd CrntEmp_state        CrntEmp_str1  CrntEmp_str2
0   (XXX, XXX)    (XXX, XXX)  (XXXX LLC, YYYY LLC)  (1234567891, 1234567899)  (12345, 12345)      (AK, AK)  (999 XXX, 999 XXX)  (XXXX, XXXX)

Upvotes: 1

mathdan
mathdan

Reputation: 191

Try forcing the data to be one column by first defining it as a Series:

import pandas as pd

pd.DataFrame(data=pd.Series(employer.values()))

Then the output matches your example desired result (which does not match your original description, but I assume you can take it from here).

                          0
0      (XXXX LLC, YYYY LLC)
1              (XXXX, XXXX)
2        (999 XXX, 999 XXX)
3                (XXX, XXX)
4            (12345, 12345)
5                (XXX, XXX)
6                  (AK, AK)
7  (1234567891, 1234567899)

Upvotes: 0

jezrael
jezrael

Reputation: 862641

I think you can use iteritems():

import pandas as pd

employer = {'CrntEmp_city': ('XXX', 'XXX'),
'CrntEmp_cntry': ('XXX', 'XXX'),
'CrntEmp_orgNm': ('XXXX LLC', 'YYYY LLC'),
'CrntEmp_orgPK': ('1234567891', '1234567899'),
'CrntEmp_postlCd': ('12345', '12345'),
'CrntEmp_state': ('AK', 'AK'),
'CrntEmp_str1': ('999 XXX', '999 XXX'),
'CrntEmp_str2': ('XXXX', 'XXXX')}

print pd.DataFrame([i for i in employer.iteritems()])
                 0                         1
0    CrntEmp_orgNm      (XXXX LLC, YYYY LLC)
1     CrntEmp_str2              (XXXX, XXXX)
2     CrntEmp_str1        (999 XXX, 999 XXX)
3    CrntEmp_cntry                (XXX, XXX)
4  CrntEmp_postlCd            (12345, 12345)
5     CrntEmp_city                (XXX, XXX)
6    CrntEmp_state                  (AK, AK)
7    CrntEmp_orgPK  (1234567891, 1234567899)

Upvotes: 1

Related Questions