Calcuating equally-spaced sums from unevenly-spaced time-series data

I need to create an equally-spaced vector of values which are the sums from a vector of unequally-spaced values.

More specifically, I have a file which has the number of bytes received and the time they were received. There are usually more than a million lines in the file. I want to plot the data rate every five minutes. To make things more difficult, the timestamps might not always be in order.

This is my code so far. I could write a loop to populate kbps, but is there an easier way to do it?

D = textscan( fid, '%f %d\n' );

datim = D{1};           % Posix timestamp
bytes = D{2};           % Number of bytes

t1 = dateshift(datetime(min(datim), 'ConvertFrom', 'posixtime'), 'start','day');
t2 = dateshift(datetime(max(datim), 'ConvertFrom', 'posixtime'), 'end','day');
t = t1:minutes(5):t2;

kbps = nan( size(t) );

And here are the variables:

[Walter Roberson]

see retime

Answer 1

[Cris LaPierre]

I would put your datetime and bytes into a table (consider cell2table). Convert your times to datetime, and then use sortrows to place them in order (if desired).

With the data in a table, you can use groupsummary to compute the number of datapoints in bins you specify, where groupbins is the bin edges.

G = groupsummary(T,groupvars,groupbins)

You could use the data you already have captured in t for your edges:

Consider sharing your data and we can test actual code. You can attach it using the paperclip icon.

[dormant]

Many thanks.

Attached is a test data file, with 20,000 lines, if you feel the need to test some code.

I will replace the textscan with readtable.

[dormant]

I just re-read your answer and think that my "many thanks" is insignificant. This is a fabulous reply. I'm quite a basic MATLAB user, but am always keen to learn the powerful things it can do.

Many, many thanks.

[dormant]

Here is my working code. Many, many, many thanks.

D = readtable( fileSniff );
D = sortrows(D,1);
D.Var1 = datetime( D.Var1, 'ConvertFrom', 'posixtime' );

tBeg = dateshift(D.Var1(1), 'start', 'hour');
tEnd = dateshift(D.Var1(end), 'end', 'hour');
T = tBeg:minutes(5):tEnd;

G = groupsummary(D,'Var1',T,'sum');

[Cris LaPierre]

Nice job and nice solution. You are very welcome.