Pandas dataframe with a tuple index and corresponding 'Num' column:
Index Num
('Total', 'A') 23
('Total', 'A', 'Pandas') 3
('Total', 'A', 'Row') 7
('Total', 'A', 'Tuple') 13
('Total', 'B') 35
('Total', 'B', 'Rows') 12
('Total', 'B', 'Two') 23
('Total', 'C') 54
('Total', 'C', 'Row') 54
Total 112
The index and 'Num' column are already sorted with a lambda function by Alphabetical Order and based on the length of tuple elements:
dataTable = dataTable.reindex(sorted(dataTable.index, key=lambda x: (not isinstance(x, tuple), x)))
Now, I want to sort only the 3rd tuple index element based on it's corresponding 'Num' value. Here would be an updated example of the dataframe:
Index Num
('Total', 'A') 23
('Total', 'A', 'Tuple') 13
('Total', 'A', 'Row') 7
('Total', 'A', 'Pandas') 3
('Total', 'B') 35
('Total', 'B', 'Two') 23
('Total', 'B', 'Rows') 12
('Total', 'C') 54
('Total', 'C', 'Row') 54
Total 112
What Lambda function can achieve this?
You can try:
def fn(x):
vals = x.sort_values(by='Num', ascending=False)
df.loc[x.index] = vals.values
m = df['Index'].apply(len).eq(3)
df[m].groupby(df.loc[m, 'Index'].str[1], group_keys=False).apply(fn)
print(df)
Prints:
Index Num
0 (Total, A) 23
1 (Total, A, Tuple) 13
2 (Total, A, Row) 7
3 (Total, A, Pandas) 3
4 (Total, B) 35
5 (Total, B, Two) 23
6 (Total, B, Rows) 12
7 (Total, C) 54
8 (Total, C, Row) 54
9 Total 112
Initial df:
Index Num
0 (Total, A) 23
1 (Total, A, Pandas) 3
2 (Total, A, Row) 7
3 (Total, A, Tuple) 13
4 (Total, B) 35
5 (Total, B, Rows) 12
6 (Total, B, Two) 23
7 (Total, C) 54
8 (Total, C, Row) 54
9 Total 112