Why can't you replace integers with lists using `replace` method - pandas

Question

So let's say i have a pandas data-frame as below:

df=pd.DataFrame({'a':[1,2,3,0]})

So my goal is to replace 0 value with [] (empty list) in this dataframe, but i did:

print(df.replace(0,[]))

But it gives me an error:

TypeError: Invalid "to_replace" type: 'int'

I tried everything that's possible i.e:

df[df==0]=[]

etc...

But nothing works.

Desired output (in case for confusion):

   a
0  1
1  2
2  3
3 []

Answer 1

It is possible by list comprehension, but because mixed content - numeric with list it is not recommended:

df['a'] = [[] if x == 0 else x for x in df.a]

print (df)

    a
0   1
1   2
2   3
3  []

And replace all values in all columns:

df = df.applymap(lambda x: [] if x == 0 else x)
print (df)
    a
0   1
1   2
2   3
3  []

Answer 2

There are two issues here. First is pandas' quirkiness when dealing with lists. To replace values in a DataFrame with list you need to do something like this;

df.loc[df.a == 0, "a"] = [[] for _ in df[df.a == 0]]

This creates n empty list based on the number of items that matches the criteria (df == 0)

The second issue is that your column is of integer type, and you can't store a list in an integer column. So before you can assign the list, you would first need to convert the column type to object first.

df = df.astype(object)
df.loc[df.a == 0, "a"] = [[] for _ in df[df.a == 0]]