CODEWITHSUNDEEP
pythonpandasdataframe
Kylo
Kylo

Reputation: 322

Replace a column value for a select set of rows

I have a pandas DataFrame for which I'd like to replace one column value for a small subset of rows. E.g. given:

in_dict = [{ 'what': 'abc', 'num': 42}, {'what': 'abc', 'num': 43}, {'what': 'def', 'num': 99}]

df = pd.DataFrame.from_records(in_dict)

which looks like:

  what  num
0  abc   42
1  abc   43
2  def   99

Using

lookup = {'abc': 'xyz'}

I want to end up with:

  what  num
0  xyz   42
1  xyz   43
2  def   99

I can use map, but then I get NaN's for entries not in the lookup table:

>>> df['what'] = df['what'].map(lookup)
>>> df
  what  num
0  xyz   42
1  xyz   43
2  NaN   99

My actual DataFrame is about 100k items, of which <100 need the update.

What is the most efficient (and ideally pandas-idiomatic) approach here?

Upvotes: 0

Views: 53

Answers (1)

user7864386
user7864386

Reputation:

You could use fillna on the mapped Series to fill it with the original:

df['what'] = df['what'].map(lookup).fillna(df['what'])

But a better option is to use replace:

df['what'] = df['what'].replace(lookup)

Output:

  what  num
0  xyz   42
1  xyz   43
2  def   99

Upvotes: 3

Related Questions