根据条件在另一个 df 的 df 子集中填充 NaN

Question

Given a reference table fallback :

    Morning     Afternoon   Evening
Red     4         6.0           13
Blue    7         NaN           9
Green   9         1.0           2

and a data.table players :

    Player  Morning     Afternoon   Evening     Team    Total
0   Bill    4.0             NaN      13.0       Red     17.0
1   Emma    NaN             NaN      NaN        Blue    0.0
2   Mike    NaN             1.0      NaN        Green   1.0
3   Jill    NaN             NaN      NaN        Red     0.0

I would like to fill NaN data in players according to the following rule: for a player missing data in all three of Morning, Afternoon, Evening (ie whose Total is zero), fill those three columns from the data in fallback matching their Team . Desired outcome:

    Player  Morning     Afternoon   Evening     Team
0   Bill    4.0            NaN      13.0        Red
1   Emma    7.0            NaN      9.0         Blue
2   Mike    NaN            1.0      NaN         Green
3   Jill    4.0            6.0      13.0        Red

Code to generate sample data:

fallback = pd.DataFrame(
    {
        'Morning': [4, 7, 9],
        'Afternoon': [6, np.NaN, 1],
        'Evening': [13, 9, 2]
    },
    index=['Red', 'Blue', 'Green'])

players = pd.DataFrame({
    'Player': ['Bill', 'Emma', 'Mike', 'Jill'],
    'Morning': [4, np.NaN, np.NaN, np.NaN],
    'Afternoon': [np.NaN, np.NaN, 1, np.NaN],
    'Evening': [13, np.NaN, np.NaN, np.NaN],
    'Team': ['Red', 'Blue', 'Green', 'Red']
})
players['Total'] = players[['Morning', 'Afternoon', 'Evening']].sum(1)

outcome = pd.DataFrame({
    'Player': ['Bill', 'Emma', 'Mike', 'Jill'],
    'Morning': [4, 7, np.NaN, 4],
    'Afternoon': [np.NaN, np.NaN, 1, 6],
    'Evening': [13, 9, np.NaN, 13],
    'Team': ['Red', 'Blue', 'Green', 'Red']
})

Answer 1

Use DataFrame.combine_first by Team created by convert column Team to index by condition - tested missing values with DataFrame.all :

df = players.set_index('Team')
m = df[['Morning','Afternoon','Evening']].isna().all(axis=1)

df[m] = df[m].combine_first(fallback)
players = df.reset_index().reindex(players.columns, axis=1)
print (players)
  Player  Morning  Afternoon  Evening   Team  Total
0   Bill      4.0        NaN     13.0    Red   17.0
1   Emma      7.0        NaN      9.0   Blue    0.0
2   Mike      NaN        1.0      NaN  Green    1.0
3   Jill      4.0        6.0     13.0    Red    0.0

Answer 2

We can do slice the with all and isna , then change the fallback to the target row index then update

player2 = player[player[['Morning','Afternoon','Evening']].isna().all(1)]
fallback = fallback.reindex(player2.Team).reset_index()
fallback.index = player2.index
player.update(fallback)