Calculate number of sprints based on sprint and time data

Hi everyone,

I am analyzing soccer-related data. I have a big file containing x & y coordinates, speed (km/h) and acceleration for each point in time (frequency was 1000 Hz but I transformed the samples into time in seconds). So, for each variable, I have its value for a lot of time-points.

I want to calculate the number of sprints performed during the match. Somehting is considered as a sprint if:

• speed > 17 km/h AND
• duration (t) > 1 sec (so speed needs to be higher than 17 km/h for an interval of minimal 1 second)

I think I need to do something with a for loop and the 'diff' function, but I'm struggling a lot... Can someone help me out? :)

Answer 1

[Aquatris]

Based on another answer here, find below a solution:

clear ,clc

Fs = 1e3; % sampling rate
dt = 1/Fs;

t = 0:dt:100; % time vector
x = sin(t)*20; % signal

threshold = 17; % signal threshold to look for
threshold_duration = 1; % min duration 
threshold_number_samples = ceil(threshold_duration * Fs); % duration in samples

%% exact copy of the solution from the link
mask = x(:).' >= threshold; % find parts of the signal that is above threshold
% create pattern of [0 1 1 1 ... 1] that defines 1s for threshold_number_samples
mask2 = true(1, threshold_number_samples);
% search for the start points for the mask2 pattern in mask to identify 
%   the occurences of signal being larger than threshold for at least some duration
starts = strfind([false, mask], [false, mask2]); 
% search for the stop points  
stops = strfind([mask, false], [mask2, false]); 

% get index of signal that is larger than threshold for at least threshold_duration
idx = []; 
for i = 1:length(starts)
    idx = [idx,starts(i):stops(i),stops(i)+(1:threshold_number_samples)];
end

% plot data and identified part of the signal that is greater than the
% threshold for at least 1 sec
plot(t,x,t(idx),x(idx),'.')

[Lotte Korblet]

Thank you some much for your help!

Unfortunately, I won't get the right answer...

My 'mask' variable is 1x168853 logical, while my speed signal is 168853x1 double. Does this matter?

Maks2 is just a 1x1000 logical variable containing only 1's, it that right?

And my 'starts' and 'stops' is 0x0 and contains nothing...

Hopefully you can help me out again :)

[Aquatris]

The code expect your signal to be a row vector. So just transpose your speed signal (or change the code to work with column vectors whichever you prefer).

mask2 should be 1x1000 for 1 sec duration so that is correct assuming your sampling time is 0.001 sec.

if your starts and stops are 0x0 in other words empty, it means there is no point in your data where the condition is true. Try to manually examine the data to see if you actually have such a case.

If it is not confidential, you can also share your data as a .mat file here and I can take a look.

[Lotte Korblet]

It worked! Thank you so much!

[Aquatris]

Glad it worked. Good luck with your work. Sounds interesting :)