How to rename columns in Pandas DataFrame in loop?

Question

I'm try to merge several dataframes in loop. My code is :

my_dict = {
        '1':{
                'providers_using':{
                        'providers':['prov1','prov2','prov3'],
                        'calls_count':[10,20,30]
                }
            },
        '2':{
                'providers_using':{
                        'providers':['prov1','prov2','prov3'],
                        'calls_count':[100,200,300]
                }
            },
        '3':{
                'providers_using':{
                        'providers':['prov1','prov2','prov3'],
                        'calls_count':[1000,2000,3000]
                }
            }

}



df_joined = pd.DataFrame(columns = ['providers','calls_count'])

for i in range(1,4):
  gw_df = pd.DataFrame.from_dict(my_dict[str(i)]['providers_using'])
  gw_df.rename(columns={"calls_count":"gw" + str(i)})
  df_joined = pd.merge(df_joined,gw_df,left_on='providers',right_on = 'providers',how='outer')

print(df_joined)

Result is:

  calls_count_x providers  calls_count_y  calls_count_x  calls_count_y
0       NaN         prov1        10             100           1000     
1       NaN         prov2        20             200           2000     
2       NaN         prov3        30             300           3000

But i want get names columns "calls_count" as gw1,gw2,gw3, etc. The count of steps in loop maybe random so it's not acceptable just do :

df_joined.columns = ['gw1','gw2','gw3']

Expected output is:

    providers    gw1            gw3           gw3
0   prov1        10             100           1000     
1   prov2        20             200           2000     
2   prov3        30             300           3000

Btw - why columns name are "calls_count_y","calls_count_x" ?

What is the expected output? Also, you have 5 columns, you cannot set 3 names only — mozway
– mozway, Commented Jul 26, 2021 at 14:05
Have you searched how to rename a single column? It seems that's the most logical option: rename column every iteration of the loop. — 9769953
– 9769953, Commented Jul 26, 2021 at 14:06
If you ask why the the _x and _y parts, that's in the documentation; see the "suffixes" parameter description. — 9769953
– 9769953, Commented Jul 26, 2021 at 15:44

harp1814 · Accepted Answer · 2021-07-26 14:31:43Z

2

Below is the trick. You can also rename the columns. The idea is to create a list of columns as per the name gw + str(i) and create a column list.

col = []
for i in range(1,4):
  col.append("gw" + str(i))
  gw_df = pd.DataFrame.from_dict(my_dict[str(i)]['providers_using'])
  df_joined = pd.merge(df_joined,gw_df,left_on='providers',right_on = 'providers',how='outer')

df_joined.columns = ['providers'] + col
print(df_joined)

Edit : As per the documentation of pandas merge, the suffixes will be _x and _y.

edited Jul 26, 2021 at 14:31

harp1814

1,6583 gold badges16 silver badges37 bronze badges

answered Jul 26, 2021 at 14:25

Shivendra Pratap Kushwaha

2,0211 gold badge11 silver badges23 bronze badges

Sign up to request clarification or add additional context in comments.

Comments

mozway · Accepted Answer · 2021-07-26 14:39:36Z

Here is a way to get the expected output from the initial dictionary without looping:

(pd.concat({k: pd.DataFrame(v).T.apply(pd.Series.explode)  # unnest the lists
            for k,v in my_dict.items()})
   .reset_index()                                          # dict keys as column
   .pivot(index='providers',                               # pivot the calls count
          columns='level_0',
          values='calls_count')
   .add_prefix('gw')                                       # rename columns
   .rename_axis('', axis=1)                                # remove 'level_0' name
   .reset_index()                                          # index as column
)

output:

  providers gw1  gw2   gw3
0     prov1  10  100  1000
1     prov2  20  200  2000
2     prov3  30  300  3000

Collectives™ on Stack Overflow

How to rename columns in Pandas DataFrame in loop?

2 Answers 2

Comments

Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

2 Answers 2

Comments

Comments

Your Answer

Sign up or log in

Post as a guest

Related