How to repeat data because of variables with two different timestamps?

I have measured data with two different devices.

One device (Garmin device) measured data every second for about 20 minutes, so I have data with a 1207x1 format.

The other device (Cosmed device) measured data every 3 seconds (so 1, 4, 7, 10, etc.) also for about 20 minutes, so this results in a 470x1 format.

I want to compare variables measured from these devices, but for a reliable analysis I have to have the same time dimension (either per 1 or 3 seconds).

So...

Therefore, I want to repeat the data for 3 seconds three times, because this does not manipulate your data with interpolation or averaging (it's the best I can do I think in this situation).

What is the simplest way to achieve this?

[dpb]

See retime with 'previous' method

[Star Strider]

Reference — Averaging data from seconds to three seconds because of different time variables

Not at all straightforward.

[Torsten]

a = [1 2 3];
b = [1 5 8 2 9 12 7 5 16];
a3 = [];
for i = 1:numel(a)
    a3 = [a3,a(i),a(i),a(i)];
end
a3

a3 = 1×9

1 1 1 2 2 2 3 3 3

Answer 1

[dpb]

Illustrate --

t=seconds(0:1:300).';                                                           % one-second time series arbitrary length
tG=timetable(t,randn(size(t)),'VariableNames',{'Garmin'});                      % create a 1-sec timetable
tC=timetable(t(1:3:end),randn(ceil(numel(t)/3),1),'VariableNames',{'Cosmed'});  % and another 3-sec
head(tC)

     Time      Cosmed 
    ______    ________

    0 sec      0.30895
    3 sec     -0.34037
    6 sec       1.6222
    9 sec      0.63342
    12 sec     -0.2297
    15 sec     -0.2933
    18 sec    -0.72891
    21 sec     0.50354

tC=retime(tC,'secondly','previous');    % convert the 3-sec to 1-sec by repeating sampled data
tGC=[tG tC];                            % now put the two together

    Time      Cosmed 
    _____    ________

    0 sec     0.30895
    1 sec     0.30895
    2 sec     0.30895
    3 sec    -0.34037
    4 sec    -0.34037
    5 sec    -0.34037
    6 sec      1.6222
    7 sec      1.6222

head(tGC)

      t       Garmin      Cosmed 
    _____    ________    ________

    0 sec    0.099782     0.30895
    1 sec     0.68994     0.30895
    2 sec      0.8777     0.30895
    3 sec      1.8498    -0.34037
    4 sec    -0.55117    -0.34037
    5 sec      1.3509    -0.34037
    6 sec    -0.35978      1.6222
    7 sec    -0.51248      1.6222

See retime for details on its features.

With two files, you'll just read each and have the time vectors in them, I presume -- so read each and augment the shorter as needed...

[Ilse Frans]

Hi, thanks so much for helping. I'm really trying to understand, but get the following error when I try tC = retime(tC, 'secondly','previous');

Incorrect number or types of inputs or outputs for function 'retime'.

[dpb]

In isolation can't tell...need see how defined tC and the context of the error; the above illustrates it works for a timetable, so yours must not have been...

What does

which -all retime

return?

And

whos tC

?

[Ilse Frans]

Found out I wasn't using timetable variables within the function 'retime' so tried making that work.

But I also found out that within the Cosmed data the time distribution is not per 3 seconds, it is unfortunately not constant.

Can I then still use the function 'retime' to let the two variables match?

[Ilse Frans]

Got to fix it using the following code:

garmin_t_new = seconds(0:1:1206).';

tG = timetable(garmin_t_new,garmin_EE_HR,garmin_VO2_HR,garmin_MET_HR,garmin_EE_GPS,garmin_VO2_GPS,garmin_MET_GPS,'VariableNames',{'EE_HR','VO2_HR','MET_HR','EE_GPS','VO2_GPS','MET_GPS'}); % create Garmin timetable

cosmed_t_new = duration(cosmed_t,'InputFormat','mm:ss');

tC = timetable(cosmed_t_new,cosmed_EEm,cosmed_VO2kg,cosmed_METs,'VariableNames',{'EEm','VO2','METs'}); % create Cosmed timetable

tC = retime(tC,garmin_t_new,'previous');

[dpb]

Yeah, retime is very flexible but it does require using a timetable to get access to it; it's understandable why is so, but would be handy if it could operate on regular datetime variables with auxiliary arrays--although it's generally not too inconvenient to just go ahead and create the needed timetable(s).