How to convert cell array to structure array

I would like to convert cell array to a structure 1x2 (or 2x1) with 3 fields with a1 as dimension/axis:
a1={'AA','AA','AA','BB','BB','BB','BB'}';
a2={'35000','96000','88000','33000','44000','8000','18000'}';
a3={'1a','1b','1c','1d','1e','1a','1b'}';
a123=[a1,a2,a3];
rowHeadings = {'first', 'second', 'third'};
but this does not work:
depts = cell2struct(a123, rowHeadings, 1)
Error using cell2struct
Unknown command option.

댓글 수: 7

Aleksander - which version of MATLAB are you using? If I try the above, on R2014a, and change the depts initialization to
depts = cell2struct(a123', rowHeadings, 1)
then it works fine, with
depts =
7x1 struct array with fields:
first
second
third
Note that the transpose of a123 is necessary since rowheadings is a 1x3 cell array, and so the first input to cell2struct must be a 3x7.
Geoff thank you - it works. However now I have 7x1 struct with 3 fields. But I should have 1x2 structure (one for AA and second for BB)as they are distinct categories with 3 fields (which is ok). That is to say I want to group AAs and BBs in 2 categories.
...a2 and a3 should be vectors instead.
Aleksander - please provide an example of how you would like the structure to appear. Also, what do you mean by a2 and a3 should be vectors instead? Aren't they already arrays?
I would like structure to appear like that: First Second Third AA [35000 96000 88000] [1a 1b 1c] BB [33000 44000 8000 18000] [1d 1e 1a 1b] The grouping categories is data in rowheadings first, i.e.: 'AA' and 'BB' (as if in idTotals in Credit risk utitilites)
Stephen23
Stephen23 2015년 6월 15일
편집: Stephen23 2015년 6월 15일
@Aleksandar: it looks like you are a bit confused about what structures can do. Structures are not really categorization tools, nor are they dictionaries (which might look similar to what you wrote above). Like most traditional MATLAB data classes the dimensions of a structure do not encode any meta-data, so you cannot "name" the dimensions of a 1x2 structure 'AA' and 'BB', atleast not as meta-data in the structure itself. It would be possible to either:
  • create a separate array containing this data, or
  • include this meta-data as data withing the structure (see my answer below).
  • use another data class: it seems like you might really be looking for something more like a table. Have you looked at using one?
Try the following:
table2struct(cell2table(.))

댓글을 달려면 로그인하십시오.

 채택된 답변

Azzi Abdelmalek
Azzi Abdelmalek 2015년 6월 13일

2 개 추천

a1={'AA','AA','AA','BB','BB','BB','BB'}';
a2={'35000','96000','88000','33000','44000','8000','18000'}';
a3={'1a','1b','1c','1d','1e','1a','1b'}';
a123=[a1,a2,a3];
rowHeadings = {'first', 'second', 'third'};
v=cell2struct(a123,rowHeadings,2)

댓글 수: 3

Hi Azzi, is there a possibility to convert it to 1x2 structure with 3 fields (AA i BB groups)?
a1={'AA','AA','AA','BB','BB','BB','BB'}';
a2={'35000','96000','88000','33000','44000','8000','18000'}';
a3={'1a','1b','1c','1d','1e','1a','1b'}';
a123=[a1,a2,a3];
rowHeadings = {'first', 'second', 'third'};
ii=unique(a1)
for k=1:numel(ii)
idx=ismember(a1,ii{k})
v(k).first=ii{k}
v(k).second=a2(idx)
v(k).third=a3(idx)
end
Note that it is poor practice in MATLAB to expand arrays inside loops, even for structures and cell arrays:
It is better either to create the entire structure in one call (see my answer), or to preallocate the structure (e.g. looping in reverse).

댓글을 달려면 로그인하십시오.

추가 답변 (1개)

Stephen23
Stephen23 2015년 6월 15일
편집: Stephen23 2015년 6월 15일

0 개 추천

Here is a short and reasonably tidy way of doing this using struct:
a1 = {'AA','AA','AA','BB','BB','BB','BB'}';
a2 = {'35000','96000','88000','33000','44000','8000','18000'}';
a3 = {'1a','1b','1c','1d','1e','1a','1b'}';
hdr = {'first', 'second', 'third'};
fun = @(x)arrayfun(@(s)x(strcmp(s,a1)), unique(a1,'stable'), 'UniformOutput',false);
X = struct(hdr{1},fun(a1), hdr{2},fun(a2), hdr{3},fun(a3));
Now lets check it in the command window:
>> X
X =
2x1 struct array with fields:
first
second
third
>> X.first
ans =
'AA'
'AA'
'AA'
ans =
'BB'
'BB'
'BB'
'BB'
>> X.second
ans =
'35000'
'96000'
'88000'
ans =
'33000'
'44000'
'8000'
'18000'

댓글 수: 2

Thank you! I accepted the answer(s) , I suppose it refers to both answers.
"I suppose it refers to both answers"
No, it doesn't. You can vote for other answers though.

댓글을 달려면 로그인하십시오.

카테고리

도움말 센터File Exchange에서 Loops and Conditional Statements에 대해 자세히 알아보기

제품

질문:

2015년 6월 13일

댓글:

2016년 7월 23일

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!

Translated by