Renaming tuple column name in dataframe

Question

I am new to python and pandas. I have attached a picture of a pandas dataframe, I need to know how I can fetch data from the last column and how to rename the last column.

Answer 1

fetch data from the last column

Retrieving the last column using df.iloc[:,-1] as suggested by other answers works fine only when it is indeed the last column.

However, using absolute column positions like -1 is not a stable solution, i.e. if you add some other column, your code will break.

A stable, generic approach

First of all, make sure all your column names are strings:

# rename columns
df.columns = [str(s) for s in df.columns]
# access column by name
df['(vehicle_id, reservation_count)']

rename the last column

It is preferable to have similar column names for all columns, without brackets in them - make your code more readable and your dataset easier to use:

# access column by name
df['vehicle_id_reservation_count`]

This is a straight forward conversion on all columns that are named by a tuple:

# rename columns
def rename(col):
    if isinstance(col, tuple):
        col = '_'.join(str(c) for c in col)
    return col
df.columns = map(rename, df.columns)

Answer 2

You can use:

df = df.rename(columns = {df.columns[-1] : 'newname'})

Or:

df.columns = df.columns[:-1].tolist() + ['new_name']

It seems solution:

df.columns.values[-1] = 'newname'

is buggy. Because after rename pandas functions return weird errors.

For fetch data from last column is possible use select by position by iloc:

s = df.iloc[:,-1]

And after rename:

s1 = df['newname']
print (s1)

Sample:

df = pd.DataFrame({'R':[7,8,9],
                   'T':[1,3,5],
                   'E':[5,3,6],
                   ('Z', 'a'):[7,4,3]})

print (df)
   E  T  R  (Z, a)
0  5  1  7       7
1  3  3  8       4
2  6  5  9       3

s = df.iloc[:,-1]
print (s)
0    7
1    4
2    3
Name: (Z, a), dtype: int64

---

df.columns = df.columns[:-1].tolist() + ['new_name']
print (df)
   E  T  R  new_name
0  5  1  7         7
1  3  3  8         4
2  6  5  9         3

---

df = df.rename(columns = {('Z', 'a') : 'newname'})
print (df)
   E  T  R  newname
0  5  1  7        7
1  3  3  8        4
2  6  5  9        3

s = df['newname']
print (s)
0    7
1    4
2    3
Name: newname, dtype: int64

df.columns.values[-1] = 'newname'
s = df['newname']
print (s)
>KeyError: 'newname'