How can i sum every 12th block of data from a large data set?

I have a set of data for energy performance sampled every 5 minutes. I would like to condense the energy outputs into hourly totals - hence blocks of 12.

I have imported the data into a variable array from Excel, called 'unit1'. the data i want lies in column 9:

>> y=(unit1(1:124390,9));

As one can see it is a large data set.

All i know so far is that i can pick out every 12th value:

>> y=(unit1(1:12:124390,9));

but that's as far as i have got - some others have done resampling but it is not clear.

In Brief - lots of data, tell the sum of every block of 12 in a continous fashion.

I then hope to produce a frequency distribution of the energy output over a period of one year (or 8760 hours)

[Image Analyst]

124390 is not a multiple of 12. What do you plan on doing with the incomplete hours? Do the first 12 data points belong to one hour, or do they span two hour bins?

[James]

I kept it simple for the question - i will use only the first complete years number of values (24x365 = 8760) so after calculating the block-sums of 12 in an new array, plot only the first 8760. the first 12 points do cover one hour. incomplete hours will be at the end and ignored.

Answer 1

[Cedric]

You can go for a solution around:

 n = size(unit1, 1) ;                   % Should be 124390,

 id = zeros(n, 1) ;                     % Build vector of ids:
 id(1:12:n) = 1 ;                       % 1 1 1 1 1 1 1 1 1 1 1 1 2 2 2 2 ...
 id = cumsum(id) ;

 sumHour = accumarray(id, unit1(:,9)) ;

[James]

Thats brilliant - thank you very much.

[Asliddin Komilov]

I need to sum variable by months so I get monthly sums:

Year=randn(1,365);

months365=[31 28 31 30 31 30 31 31 30 31 30 31];

any help is appreciated.

thanks

Answer 2

[Image Analyst]

Here it is for data that is a multiple of 12 long:

% Create sample data a multiple of 12 long.
unit1 = rand(124392, 9);
% Exact column #9.
column9 = unit1(:,9);
% Reshape it into a 10366 row by 12 column matrix.
reshapedColumn9 = reshape(column9, [length(column9)/12, 12]);
% Get the means going cross columns.
theMeans = mean(reshapedColumn9, 2);
% These are the means of blocks of 12 in the original matrix.

Of course someone will pack that all into a single line but I made it explicitly in several lines and put in comments so that you can follow the thought process more easily.

[Cedric]

I'm voting for this solution; I proposed ACCUMARRAY before the OP wrote the comment indicating that he would truncate the data to a multiple of 12, and CUMSUM/ACCUMARRAY were a solution for managing situations where the "multiple of 12" condition is not necessary.