How can I include absent group names in table grouping results?

Question

Aaron Thrasher 2021-10-15

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/1564656-how-can-i-include-absent-group-names-in-table-grouping-results

评论： Aaron Thrasher 2021-10-15

I am using groupsummary on table information and am finding it overly complicated to show additional data in the results. Here is an example of my situation:

teamMembers = {'Al','Betty','Charles','Denice','Edward'};
%Example Data
Name = {'Denice';'Denice';'Al';'Charles';'Al';'Denice'};
Month = [1;1;2;2;2;2];
Hours = [1;3;3;2;4;2];
%Table
T = table(Name,Month,Hours);
%Example summary (current result)
weekHours = groupsummary(T,{'Month','Name'},'sum','Hours')
weekHours = 4×4 table
    Month       Name        GroupCount    sum_Hours
    _____    ___________    __________    _________

      1      {'Denice' }        2             4    
      2      {'Al'     }        2             7    
      2      {'Charles'}        1             2    
      2      {'Denice' }        1             2    
%Example summary (expected result)

Month Name GroupCount sum_Hours

1 {'Al' } 0 0

1 {'Betty' } 0 0

1 {'Charles'} 0 0

1 {'Denice' } 2 4

1 {'Edward'} 0 0

2 {'Al' } 2 7

2 {'Betty' } 0 0

2 {'Charles'} 1 2

2 {'Denice' } 1 2

2 {'Edward'} 0 0

But I would like to include empty groups of the team as in the example for categorical information using 'IncludeEmptyGroups'

In the end I'd like all team members to show up in the summary which would show group count and sum of 0.

So far I have tried just inlcuding these names as an extra row in the table, but this is much more complicated in a real example with a growing number of table columns and data types to append the table with empty data just to include the names. I could always add the empty names at the end summary with 0's, I'm just looking for a more convenient way to do this. I haven't explored categorical types so I am not sure if there is a better solution there either.

0 个评论
显示 -2更早的评论隐藏 -2更早的评论

请先登录，再进行评论。

请先登录，再回答此问题。

Answer 1

dpb 2021-10-15

0
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/1564656-how-can-i-include-absent-group-names-in-table-grouping-results#answer_809516

Well, unless the data are in the table, there's not going to be anything for groupsummary to operate over to tell it those other elements even exist. Don't see much of any way around that if you want groupsummary to do all the work.

Otherwise you could just take the results you get and augment the summary table with the missing names, but that function will have to know which variables are in the original in order to add the proper variable(s) to the summary table.

I don't see it as being that hard to create an empty record of the data type as the table and making the entries.

0 个评论
显示 -2更早的评论隐藏 -2更早的评论

请先登录，再进行评论。

Answer 2

Image Analyst 2021-10-15

0
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/1564656-how-can-i-include-absent-group-names-in-table-grouping-results#answer_809556

Try this:

teamMembers = {'Al','Betty','Charles','Denice','Edward'}
%Example Data
Name = {'Denice';'Denice';'Al';'Charles';'Al';'Denice'}
Hours = [1;3;3;2;4;2];
% Add in missing names
missingNames = setdiff(teamMembers, Name)'
% Add zeros for them to the lists
Name = [Name; missingNames]
Hours = [Hours; zeros(numel(missingNames), 1)]
%Table
T = table(Name,Hours);
%Example summary (current result)
weekHours = groupsummary(T,'Name','sum','Hours')
%Example summary (expected result)

weekHours =

5×3 table

Name GroupCount sum_Hours

___________ __________ _________

{'Al' } 2 7

{'Betty' } 1 0

{'Charles'} 1 2

{'Denice' } 3 6

{'Edward' } 1 0

2 个评论
显示无隐藏无

dpb 2021-10-15

Nice demo of using setdiff to get the missing, @Image Analyst.

I'd point out that one could use vartype and/or setvartype and friends to build a record that matches a given table structure programmatically as well as just having a set of numeric variables so that the additional code could be dynamic as/if the table content changes.

Aaron Thrasher 2021-10-15

I explored a little bit of using types to add new data like you described, however it got pretty complicated when trying to set the variable type, but this was without the setvartype, so this could be very useful. In the end my current solution is to have a cell array displaying the data I need with all blank answers, then using intersect from the group summary to place in the data. With having database queries of useful tables with a width more than 20 columns and multiple datapoints and up to 3 level groupings, it gets complicated quick. Thanks for all your help everyone.

请先登录，再进行评论。