0
data = [{'a': 12, 'b': 2, 'c': 3, 'd': 'bat'},
        {'a': 'NaN', 'b': 20, 'c': 30, 'd': 'ball'},
        {'a': 4, 'b': 20, 'c': 30, 'd': 'pin'}]

df = pd.DataFrame(data)

I'm having a hard time figuring out how to replace the NaN values in column A with values in column b based on conditions set on columns c and d. For example, if I wanted to replace the NaN values in the a column with the values of column b (2 and 20, respectively) under circumstances where the value of c > 20 and 'd' = 'ball.

Could someone help me with this?

I've tried a number of solutions with df.loc and df.mask that have not worked.

Aleisha Khan
  • 3
  • 2
  • Does this help: https://stackoverflow.com/questions/55081194/replacing-values-in-a-pandas-dataframe-based-on-multiple-conditions – Barmar Jul 26 '23 at 19:47
  • Unfortunately, no. The issue is that I am trying to replace values in column a with corresponding row values from column b. I am not simply trying to replace all NaN values with a particular number or string, as many of these examples show. – Aleisha Khan Jul 26 '23 at 19:55
  • You can use `df['other_column']` as the source value instead of a particular number. – Barmar Jul 26 '23 at 19:56

2 Answers2

0

You can use df.apply for this

def fill_na(row: pd.Series):
    if pd.isna(row[0]):
        if row[2] > 20 and row[3] == "ball":
            row[0] = row[1]
    return row

df = df.apply(fill_na, axis=1)

As you don't have rows that would satisfy your criteria, you can use this for testing

[{'a': np.NaN, 'b': 2, 'c': 3, 'd': 'bat'},
 {'a': 10, 'b': 20, 'c': 30, 'd': 'ball'},
 {'a': np.NaN, 'b': 20, 'c': 30, 'd': 'pin'},
 {'a': np.NaN, 'b': 15, 'c': 30, 'd': 'ball'}]
Andrey
  • 340
  • 2
  • 11
0

Try:

df['a'].loc[(df['a'].isna())&(df['c'] > 20) & (df['d'] == 'ball')] =df['b'].loc[(df['a'].isna())&(df['c'] > 20) & (df['d'] == 'ball')]
jjk
  • 111
  • 9
  • When I try this, it I get the warning: value is trying to be set on a copy of a slice from a DataFrame. I need the change to be applied to the original dataframe. – Aleisha Khan Jul 26 '23 at 20:27
  • Answer needs supporting information Your answer could be improved with additional supporting information. Please [edit] to add further details, such as citations or documentation, so that others can confirm that your answer is correct. You can find more information on how to write good answers [in the help center](https://stackoverflow.com/help/how-to-answer). – moken Jul 31 '23 at 04:29