Pandas GroupedBy Dataframe 按列值排序

Question

My question is about sorting. I've read here multiple questions about groupby dataframes sorting but none of them solved my problem or maybe I'm doing something wrong because I get different errors like bool not callable or sort_values not available or stuff like that.

I got a dataframe with some info and columns. Then I created a groupby dataframe properly, based on 2 fields.

Then I do this:

for name, group in mydfgrouped:
    print(name, len(group))

The output I get is this:

('A', 'HH') 414
('A', 'LW') 1413
('A', 'ME') 1458
('A', 'MR') 339
('B', 'HH') 126
('B', 'LW') 288
('B', 'ME') 315
('B', 'MR') 54
('C', 'HH') 879
('C', 'LW') 672
('C', 'ME') 984
('C', 'MR') 186
('D', 'HH') 246
('D', 'LW') 756
('D', 'ME') 795
('D', 'MR') 255

I would like to sort this dataframe based on 2 criteria:

1. len of group HH descending order
2. Name ascending order (this is just in case there's a tie in 'HH')

So the idea is to show the same list but descending sorted depending on the value of HH , but showing all stats for that name.

My expected output is:

('C', 'HH') 879
('C', 'LW') 672
('C', 'ME') 984
('C', 'MR') 186
('A', 'HH') 414
('A', 'LW') 1413
('A', 'ME') 1458
('A', 'MR') 339
('D', 'HH') 246
('D', 'LW') 756
('D', 'ME') 795
('D', 'MR') 255
('B', 'HH') 126
('B', 'LW') 288
('B', 'ME') 315
('B', 'MR') 54

As you can see, name C is first because it got the highest len in group HH (879). And I want to get all groups from C . The last one is B because it got the lowest len of group HH (126)

sort_values and sort did not work for me

Answer 1

Assuming the second column that you're grouping by is called y , you can do:

# mydfgrouped = df.groupby(['x', 'y'])

s = mydfgrouped.size().unstack('y').sort_values('HH', ascending=False).stack()

for (i, l) in s.items():
    print(i, l)

Output:

('C', 'HH') 879
('C', 'LW') 672
('C', 'ME') 984
('C', 'MR') 186
('A', 'HH') 414
('A', 'LW') 1413
('A', 'ME') 1458
('A', 'MR') 339
('D', 'HH') 246
('D', 'LW') 756
('D', 'ME') 795
('D', 'MR') 255
('B', 'HH') 126
('B', 'LW') 288
('B', 'ME') 315
('B', 'MR') 54

Brief explanation:

# unstack `y` into columns
mydfgrouped.size().unstack('y')

# y   HH    LW    ME   MR
# x                      
# A  414  1413  1458  339
# B  126   288   315   54
# C  879   672   984  186
# D  246   756   795  255

# sort by HH
mydfgrouped.size().unstack('y').sort_values('HH', ascending=False)

# y   HH    LW    ME   MR
# x                      
# C  879   672   984  186
# A  414  1413  1458  339
# D  246   756   795  255
# B  126   288   315   54

# stack `y` back into rows
mydfgrouped.size().unstack('y').sort_values('HH', ascending=False).stack()