subsetting dates in a matrix

Hello,

I have a time series in a matrix and first column is serial date numbers (up to millisecond precision), and other columns my samples. I would like to get those rows that belong to for example '12-Oct-2012'. I couldn't figure out how to write the logical operator to get a slice of this matrix. I would like to hopefully avoid some sort of slow string comparison.

Thanks in advance

[Azzi Abdelmalek]

Show the two first rows of your data

[Cagdas Ozgenc]

735236.4590277778  1.4264e+03  1.4272e+03  1.4217e+03
735236.4597222222  1.4237e+03  1.4252e+03  1.4234e+03
735236.4604166667  1.4247e+03  1.4267e+03  1.4244e+03

Answer 1

[Cedric]

You already have serialized/numeric time stamps in you array, so just convert date boundaries to numeric, and compare numbers: assuming your array is named data..

 lb_incl = datenum( '12-Oct-2012', 'dd-mmm-yyyy' ) ;
 ub_excl = datenum( '13-Oct-2012', 'dd-mmm-yyyy' ) ;
 id = data(:,1) >= lb_incl & data(:,1) < ub_excl ;
 dailyMean = mean( data(id,2) ) ;

or

 dailySlice = data(id,2:end) ;

[Cagdas Ozgenc]

Thanks. In this scenario this is a good option.

[Cedric]

You're welcome.

Answer 2

[Azzi Abdelmalek]

If A is your cell array

c1=cellstr(datestr(A(:,1),'dd-mm-yyyy'))
idx=ismember(c1,'12-10-2012')
out=A(idx,:)

[Cagdas Ozgenc]

Takes forever on my dataset of million lines. It converts all serial numbers back to string and then compares those strings. Then gets the indices. Then gets the data.

Thanks anyways