Delimate Table and Rewrite it

Question

Seum Bin Rahman 2019-10-13

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/484976-delimate-table-and-rewrite-it

评论： Peter Perkins 2019-10-15

Service.xlsx

I have a Table. One of the colums has multiple strings seperated by comma. The table is as follows;

            MODELS                           PROBLEMS                  
    _______________________    ____________________________________ 
    'Olvio L25'                'CHARGING,keypad lock,DEAD'                               
    'Olvio L26'                'DEAD,Auto ON OFF'                                     
    'Olvio L26'                'KEYPAD'

I wish to delimate column "PROBLEMS" and save it back to another table which will have three columns, POBLEMS1, PROBLEMS2, PROBLEMS3.

The code is as follows

    raw=readtable('Service.xlsx');
r=0;
raw_problem=raw.Expert_Found_Problems;
  
P=height(raw);
for i=1:1:P
  PROBLEMS=raw_problem{i,1};
  C = strsplit(PROBLEMS,',');
   r=r+1
  
    T(r,:)=table(C)
end
T=T

Error is as follows

Error using t3 (line 15)
Subscripted assignment dimension mismatch for table variable 'C'.

Excel file is also attached

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

Walter Roberson 2019-10-13

0
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/484976-delimate-table-and-rewrite-it#answer_396095

table() applied to the results of strsplit is probably giving you a table with one variable and three rows.

You need to take special steps when the input is a vector because there is an ambiguity over whether to create a table with one row and several variables, or a table with several rows and one variable.

table(C{1},C{2},C{3}, 'VariableNames', {'Problem1', 'Problem2', 'Problem3'})

5 个评论
显示 3更早的评论隐藏 3更早的评论

Walter Roberson 2019-10-13

In MATLAB, there is no direct (and readable) way to call a function and immediately select a subset of the result. It would be nice if it were permitted to do something like,

[V{:}, {'','',''}](1:3) %not actually permitted

but MATLAB does not provide for that. It does, though, make it possible to call a function on a computed result, and for the function to take the subset.

Thus,

Pad3 = @(V) First3([V{:}, {'','',''}]);

will be passed a 1 x something cell array with a varying number of elements in it, from 0 to 3. The {:} part expands the cell into individual arguments of the [] function it is inside; that will expand to 0, 1, 2, or 3 arguments. Then the cell with 3 empty strings is appended through the [] operator, which gives you a cell result with 3, 4, 5, or 6 elements, with the last 3 being empty in each case. Then First3 is called on that, returning the first 3 of the elements. So {} the empty cell would become {'' '' ''} and {'one'} would become {'one' '' ''} and {'one' 'two'} would become {'one' 'two' ''} and {'one' 'two' 'three'} would stay {'one' 'two' 'three'}.

The raw_problem = raw.Expert_Found_Problems line is going to give you a cell array column vector each containing a character vector. The C = regexp(raw_problem, ',', 'split') line is going to give you a cell array the same size, with each entry being a cell array that has as many entries as there are comma deliminated parts. Like,

{
 {'CHARGING' 'keypad lock' 'DEAD'}
 {'DEAD' 'AUTO on off'}
 {'KEYPAD'}
 }

The arrayfun() is going to process each of those rows, so it is going to pass PAD3 {'CHARGING' 'keypad lock' 'DEAD'} first, then {'DEAD' 'AUTO on off'} then {'KEYPAD'} . As indicated above, that will convert to

{
  {'CHARGING' 'keypad lock' 'DEAD'}
  {'DEAD' 'AUTO on off' ''}
  {'KEYPAD' '' ''}
}

Notice that all of the entries now contain exactly the same number of columns.

Now that they all contain the same number of columns, the line

C = vertcat(C{:});

converts that to

{
  'CHARGING' 'keypad lock' 'DEAD'
  'DEAD' 'AUTO on off' ''
  'KEYPAD' '' ''
}

which is a something-by-3 cell array.

With that in place, cell2table() constructs a table with 3 variables (one for each column).

Peter Perkins 2019-10-15

In recent versions of MATLAB, you could also overwrite raw.PROBLEMS (or whatever it's called, I can't tell) with Walter's C, a "something-by-3 cell array", and then use splitvars on raw to split it into three separate variables.

请先登录，再进行评论。