Converting 'no' and 'yes' into 0 and 1 in pandas dataframe

Question

I want to convert data of 'edjefe' column which contains int as well as 'yes' and 'no' values. My problem is I just want to map 'yes' and 'no' to 1 and 0 and keep the int values as it is So I wrote this code

def foo(x):
    if x == 'no':
        return 0
    elif x == 'yes':
        return 1
    else:
        return x

and df1.edjefe.map(lambda x : foo(x))

But I am getting an error as,

RecursionError: maximum recursion depth exceeded while calling a Python object

maybe `apply` the function instead of `map`. But hard to assess the error without a working example — Bryce Ramgovind, Aug 03 '18 at 12:36
Question has nothing to do with `machine-learning` (or `numpy`) - kindly do not spam the tags (removed) — desertnaut, Aug 03 '18 at 12:36

score 17 · Accepted Answer · answered Aug 03 '18 at 12:38

17

You can also just use replace:

df.edjefe.replace(to_replace=['no', 'yes'], value=[0, 1])

answered Aug 03 '18 at 12:38

ksbg

3,214
1
22
35

`TypeError: Cannot compare types 'ndarray(dtype=int64)' and 'str'` – Aptha Gowda Aug 03 '18 at 12:40
String replace requires a string as an input and an output. Replace yes and no with `value=['0', '1']`, then cast the result with `int()`. – addohm Aug 03 '18 at 12:47
Note also that `replace` has a flag `inplace=True` for replacing the values in the dataframe in-place, rather than returning the modified values. – Anton Schwaighofer Nov 21 '22 at 11:27

score 8 · Answer 2 · answered Aug 03 '18 at 12:46

8

Just use dict-like to_replace:

df['edjefe'].replace({'no': 0, 'yes': 1})

answered Aug 03 '18 at 12:46

Lev Zakharov

2,409
1
10
24

what about all the columns? – mLstudent33 Mar 13 '20 at 00:39
1

@mLstudent33, just use `replace` on the full DataFrame, not specific column: `df.replace({'no': 0, 'yes': 1})`. – Lev Zakharov Mar 13 '20 at 08:18

jpp · Answer 3 · 2018-08-03T12:46:53.653

4

You can use pd.Series.map with a dictionary mapping followed by pd.Series.fillna:

d = {'no': 0, 'yes': 1}
df1['edjefe'] = df1['edjefe'].map(d).fillna(df1['edjefe'])

You will likely find this more efficient than pd.Series.replace.

See Replace values in a pandas series via dictionary efficiently for more details.

If you have mutable objects in your series, this will fail, since dictionary keys must be hashable. You can convert to strings in this case:

df1['edjefe'] = df1['edjefe'].astype(str).map(d).fillna(df1['edjefe'])

edited Aug 03 '18 at 12:46

answered Aug 03 '18 at 12:39

jpp

159,742
34
281
339

I am getting `TypeError: 'Series' objects are mutable, thus they cannot be hashed` error – Aptha Gowda Aug 03 '18 at 12:43
@JaqenH'ghar, Can you provide a [mcve]? I cannot replicate your error. – jpp Aug 03 '18 at 12:44
@JaqenH'ghar, Also, see update. Seems you might have `list` or other mutable types in your series. – jpp Aug 03 '18 at 12:47
Got it the 0 and 1 are of type str. Tnx – Aptha Gowda Aug 03 '18 at 13:06
@JaqenH'ghar, Nope that's incorrect. We're converting series elements to `str` as an intermediary step to ensure elements are hashable. – jpp Aug 03 '18 at 13:14

score 1 · Answer 4 · answered Apr 07 '20 at 08:55

1

You can also try:

df1['edjefe'] = (df1['edjefe']=="yes")*1

answered Apr 07 '20 at 08:55

Şenol Kurt

31
4

score 0 · Answer 5 · answered May 19 '20 at 14:32

0

You can use pandas.Categorical as well.

df1["edjefe"] = pd.Categorical(df1["edjefe"]).codes

Visit here for more information.

answered May 19 '20 at 14:32

Maryam Bahrami

1,056
9
18

Converting 'no' and 'yes' into 0 and 1 in pandas dataframe

5 Answers5