Replacing values with groupby means

Question

I have a DataFrame with a column that has some bad data with various negative values. I would like to replace values < 0 with the mean of the group that they are in.

For missing values as NAs, I would do:

data = df.groupby(['GroupID']).column
data.transform(lambda x: x.fillna(x.mean()))

But how to do this operation on a condition like x < 0?

Thanks!

Answer 1

Here's one way to do it (for the 'b' column, in this boring example):

In [1]: df = pd.DataFrame([[1,1],[1,-1],[2,1],[2,2]], columns=list('ab'))
In [2]: df
Out[2]: 
   a  b
0  1  1
1  1 -1
2  2  1
3  2  2

Replace those negative values with NaN, and then calculate the mean (b) in each group:

In [3]: df['b'] = df.b.apply(lambda x: x if x>=0 else pd.np.nan)
In [4]: m = df.groupby('a').mean().b

Then use apply across each row, to replace each NaN with its groups mean:

In [5]: df['b'] = df.apply(lambda row: m[row['a']]
                                       if pd.isnull(row['b'])
                                       else row['b'],
                           axis=1) 
In [6]: df
Out[6]: 
   a  b
0  1  1
1  1  1
2  2  1
3  2  2

Answer 2

Using @AndyHayden's example, you could use groupby/transform with replace:

df = pd.DataFrame([[1,1],[1,-1],[2,1],[2,2]], columns=list('ab'))
print(df)
#    a  b
# 0  1  1
# 1  1 -1
# 2  2  1
# 3  2  2

data = df.groupby(['a'])
def replace(group):
    mask = group<0
    # Select those values where it is < 0, and replace
    # them with the mean of the values which are not < 0.
    group[mask] = group[~mask].mean()
    return group
print(data.transform(replace))
#    b
# 0  1
# 1  1
# 2  1
# 3  2