从第二列替换列中的 NaN 值

Question

I would like to replace NaN values in Target with the corresponding Node value. My data is:

 Node  Target   Color
node1   node7   Red
node1   node9   Red
node3   node5   Green
node1   node3   Red
node3   node1   Red
node5   NaN     Yellow

I would need to have:

   Node  Target     Color
    node1   node7   Red
    node1   node9   Red
    node3   node5   Green
    node1   node3   Red
    node3   node1   Red
    node5   node5   Yellow # here the replacement

I think that a possible solution could be using an if statement to check if a node has Target equal to NaN: if yes, then it would be possible to assign itself as target.

Answer 1

Yes, df.fillna(value, ...) will allow the value (replacement) arg to be a Series (column) , not just a constant:

df['Target'] = df['Target'].fillna(df['Node'])

Note this is better than if...else logic because it does one vectorized assignment to the entire dataframe, as the testcase below shows.

---

2. Alternative, if df.fillna() hadn't allowed us to do this:

You could also use df.where() on your column of interest, where the df.where(... other) arg is your replacement column, df['Node'] :

df['Target'] = df['Target'].where(df['Target'].notna(), df['Node'])

Note also how we use logical indexing, using df['Target'].notna() to get us a mask/ logical vector of rows where Target is/isn't NA.

Better testcase:

import pandas as pd
from io import StringIO

df = """ Node  Target   Color
node1   node7   Red
node1   node9   Red
node2   NaN     Brown
node3   node5   Green
node1   node3   Red
node3   node1   Red
node5   NaN     Yellow"""

df = pd.read_csv(StringIO(df), sep=r'\s+')