Reputation: 133
Is there a canonical way to compute the element-wise mean of a list of DataFrames with identical columns and indices?
The best way I can think of is
from functools import reduce
dfs = [df1, df2, df3, df4, df5]
reduce(lambda x, y: x.add(y), dfs) / len(dfs)
Upvotes: 3
Views: 1467
Reputation: 863801
Use concat
with mean
per index values:
df1 = pd.DataFrame({
'C':[7,8,9],
'D':[1,3,5],
})
df2 = pd.DataFrame({
'C':[4,2,3],
'D':[7,1,0],
})
df3 = pd.DataFrame({
'C':[9,4,2],
'D':[1,7,1],
})
from functools import reduce
dfs = [df1, df2, df3]
df = reduce(lambda x, y: x.add(y), dfs) / len(dfs)
print (df)
C D
0 6.666667 3.000000
1 4.666667 3.666667
2 4.666667 2.000000
df = pd.concat(dfs).mean(level=0)
print (df)
C D
0 6.666667 3.000000
1 4.666667 3.666667
2 4.666667 2.000000
Upvotes: 4