# selection of data from matrix

33 views (last 30 days)
MUKESH KUMAR on 17 Nov 2017
Edited: MUKESH KUMAR on 18 Nov 2017
I have a dataset lets say a matrix of 100 rows and 5 columns, each row represents a dataset.Now I want to select 5 any rows(datasets) initially, then further I want to select 4 rows(dataset) except initially 5 rows should not be repeated, then next I want to select 10 rows except for initially selected rows (those 9 rows should not be repeated here) in before and so on..... one thing to remember that one rows should be selected once here.

Andrei Bobrov on 17 Nov 2017
data = randi([-41 78],100,5);
ii = randperm(size(data,1));
k = [5 4 10];
out = mat2cell(data(ii(1:sum(k)),:),k,size(data,2));
MUKESH KUMAR on 18 Nov 2017
Edited: MUKESH KUMAR on 18 Nov 2017

KL on 17 Nov 2017
Edited: KL on 18 Nov 2017
One easy alternative is to shuffle the row of your matrix first and then extract them from top to bottom,
data = [(1:100).' rand(100,5)]; %dummy data
indx = randperm(size(data,1));
data = data(indx,:);
numberofRowsWanted = 4;
data_ext = data(1:numberofRowsWanted,:);
Another alternative is to split and put them all in a cell array.
data = [(1:100).' rand(100,5)]; %dummy data
indx = randperm(size(data,1)); %generate random indices without repeating
splits = [4,5,10]; %how many rows you want to extraxt each time, extend it as you wish
%now, use arrayfun to extract those rows and store them in a cell array.
data_ext = arrayfun(@(x,y) data(indx(x:y),:),[1 splits+1],[splits(1) splits(1:end-1)+splits(2:end)],'uni',0);

KSSV on 17 Nov 2017
Check the below example code:
K = 1:100 ;
pick = [5 4 9 10] ;
for i = 1:length(pick)
take = randsample(K,pick(i))
% remove the selected numbers
K = setdiff(K,take) ;
end
##### 2 CommentsShow 1 older commentHide 1 older comment
KL on 18 Nov 2017
It indeed works, in fact all of the solutions here do what you want, they are just examples. Instead of just copy-pasting, try and understand how it works.

Guillaume on 17 Nov 2017
There are many ways you could do this. You could keep track of the row you've selected and exclude them from the set of rows to select from, but probably the simplest way would be to remove the rows you've selected from your dataset.
dataset = rand(100, 5); %demo dataset
while ~isempty(dataset)
%select 5 rows at random:
numrowstoselect = 5; %adjust as required
selectedrows = randperm(size(dataset, 1), min(numrowstoselect, size(dataset, 1));
selecteddataset = dataset(selectedrows, :);
%remove selected rows from dataset so they can't be selected again
dataset(selectedrows, :) = [];
end