Asked by Nathan
on 23 May 2013

Hi I have a matrix that is 43885x1. The matrix is composed of dates from 1/20/2009 to 2/20/2009. The matrix flows in numerically increasing order but with duplicates for each date. I want to group each day into its own variable with corresponding indicies. So for a matrix for 1/20/2009 would be all the indicies that originally were 1/20/2009 from the original large matrix, i.e.

X1=[ 1 2 3 ... 8567]

And for 2/20/2009 it would read something like

X32=[ 37568 37569 37570 .... 43885]

Answer by the cyclist
on 23 May 2013

Edited by the cyclist
on 23 May 2013

Accepted answer

Here is an example:

D = {'01/01/2001';'01/01/2001';'01/01/2001';'02/02/2002';'03/03/2003';'03/03/2003'};

[i,j,k] = unique(D);

numberDates = numel(i);

X = cell(numberDates,1);

for nd=1:numberDates

X{nd} = find(k==nd);

end

Notice that I used a single cell array X{1}, X{2}, etc. to store your output instead of separate variables X1, X2, etc. You will find this is a much more workable approach.

Answer by Sean de Wolski
on 23 May 2013

Hi Nathan,

Here is the general workflow you will want to use:

- Use
`datenum()`to convert the dates to serial date numbers - Then, use
`unique()`on this while capturing the first and thirs output:

[uv,~,idx] = unique(X);

- uv will contain the unique values and idx will be the size of x with each of its values pointing to the corresponding unique value.

From here it depends on what you want to do with this information. You have the indices, idx, where each index represents the value. You can use these as the *subs* in `accumarray` to perform a function on some other data. A better detailed example with all inputs, and expected outputs would help us help you with this.

Nathan
on 23 May 2013

So basically what I have here are 30 spreadsheets each in a similar format. 5 pairs of columns of a timestamp and 5 readings. The temperatures are from different meters and record at the same time. Then theres a sixth pair of coulums of a timestamp and a dew point based on these temperatures.

We first have to seperate each timestamp column by day and get the corressponding temps foreach index.

The goal in the end is to find when the dew point reaches a certain degrees C from as T increases during a heating cycle. And also find when the dew point reaches a different certain degrees C on the cooling cycle. And find the corresponding readings at this point from the other 5 meters.

Answer by David Sanchez
on 23 May 2013

A=[ 1;1;2;2;2;3;3;3;4;4;4]; % example matrix [B ind]=unique(A); L = length(ind); C = cell(L,1); C{1} = A(1:ind(1)); for k = 2:L C{k} = A( (ind(k-1)+1):ind(k) ); end

You will end up with a cell array whose cells contain a matrix for each date

Opportunities for recent engineering grads.

## 5 Comments

## Iain (view profile)

Direct link to this comment:http://www.mathworks.com/matlabcentral/answers/76771#comment_150569

I think "unique" might have the outputs you want, though, it might not be in the form you need.

## the cyclist (view profile)

Direct link to this comment:http://www.mathworks.com/matlabcentral/answers/76771#comment_150570

Iain, I suggest you put your answer as an answer, and not as a comment.

## the cyclist (view profile)

Direct link to this comment:http://www.mathworks.com/matlabcentral/answers/76771#comment_150572

Nathan, how is your original matrix stored? Are the dates strings in a cell array, or datenums in a numerical array?

## Iain (view profile)

Direct link to this comment:http://www.mathworks.com/matlabcentral/answers/76771#comment_150574

I did that earlier with a not-fully-explicit answer and got told off for making it an answer...

## the cyclist (view profile)

Direct link to this comment:http://www.mathworks.com/matlabcentral/answers/76771#comment_150578

Hm. OK, I can see that point of view, too. Sorry you got "told off", though!