How to speed up data processing when extracting from a large cell array?

5 次查看（过去 30 天）

Raju Kumar 2023-9-19

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/2022987-how-to-speed-up-data-processing-when-extracting-from-a-large-cell-array

评论： Dyuman Joshi 2023-9-19

array.mat

I am dealing with a large cell array (e.g., 3826341x1 cell). I would like merge the data or matrix from each cell. Using 'cell2mat' takes so much of time. Is there any alternative to cell2mat to process the data faster? Any leads will be highly appreciated. Thanks.

5 个评论
显示 3更早的评论隐藏 3更早的评论

Raju Kumar 2023-9-19

Hi @Dyuman Joshi, I have attached a file of small size. The actaul file is a way bigger, but can not be uploadded here because Matlab allows only up to 5MB.

Dyuman Joshi 2023-9-19

在 MATLAB Online 中打开

array.mat

With the inbuilt functions, vertcat is the fastest option.

load('array.mat')
whos
  Name                     Size               Bytes  Class    Attributes

  alphaClusterAll      14879x1             24311128  cell               
  ans                      1x34                  68  char               
  cmdout                   1x33                  66  char               
f1=@(x) cell2mat(x);
f2=@(x) vertcat(x{:});
f3=@(x) cat(1,x{:});
F1 = @()f1(alphaClusterAll);
F2 = @()f2(alphaClusterAll);
F3 = @()f3(alphaClusterAll);
fprintf('Time taken by cell2mat = %f seconds', timeit(F1))
Time taken by cell2mat = 0.009412 seconds
fprintf('Time taken by vertcat = %f seconds', timeit(F2))
Time taken by vertcat = 0.003332 seconds
fprintf('Time taken by cat = %f seconds', timeit(F3))
Time taken by cat = 0.005746 seconds
isequal(F1(),F2(),F3())
ans = logical
   1