Set all the values in multiple columns to empty lists pandas-CodePudding

I have a dataframe which have lists or values in their columns; something like the following:

df
     A        B        C      D
0   []      [3]    ['ON']     5
1   'a'   ['a']    ['ON']     5
2    5      [3]    ['ON']     5
3   []      [3]    ['ON']     5
...

I would like to replace all the values inside columns A, B, and C with empty lists. I tried using .assign(column_name='value') seperatly for the columns A, B, and C. I can set a value but I cannot set an empty list. I do not want to use .apply(lambda x: []), since it is rather slow.

Is there any other way?

Expected Outcome:

df
     A    B    C    D
0   []   []   []    5
1   []   []   []    5
2   []   []   []    5
3   []   []   []    5
...

what I basically need isa pandas function which can do: change everything in columns=['A','B','C'] to []

CodePudding user response：

You can use:

df['A'] = [[]]*len(df)

CodePudding user response：

Try setting the column with a list comprehension.

E.g.

empty_col = [[] for x in range(len(df))]
df['A'] = empty_col
df['B'] = empty_col
df['C'] = empty_col

>>> df
         A    B    C    D
    0   []   []   []    5
    1   []   []   []    5
    2   []   []   []    5
    3   []   []   []    5
...

CodePudding user response：

df['A'] = [np.empty(0,dtype=float)]*len(df)
df['B'] = [np.empty(0,dtype=float)]*len(df)
df['C'] = [np.empty(0,dtype=float)]*len(df)

Performance comparison:

for seed data:

df = pd.DataFrame(columns=['A'])
for i in range(100):
    df = df.append({'A': i}, ignore_index=True)
df

With 1 000 elements: 396 µs vs 613 µs

With 10 000 elements: 1.06 ms vs 4.33 ms

With 100 000 elements: 8.87 ms vs 45.9 ms