Calculate the duration that variable remains in a specific value

Hello,

i need to calucale, how long are the intervals a variable with a specific value stays until it change again. In the picture below you can see the plot of the variable i am working with (speed vs time in min).

i would like to have a result like this: The Speed 3000 have followed Interval duration in the Dataset

S_3000 = [6 5 4 7 8 9 4 5] % in Minutes or Datapoints

Where the entries of S_3000 are the number of Intervals found in the Dataset. I have tried already some Ideas, but i don't find any solution. I will also attacht the Data i am working with. Thanks in Advance!

Answer 1

[Kevin Holly]

load('tab.mat')

logical_array = tab.s==3000; %chose value for speed here
bar(tab.t,logical_array)

t2 = table;
count = 0;
timepoint = "";
epochtime = [];

Here, I detected when the speed stayed at the value of 3000 for at least 3 datapoints. In addition to calculating the duration I also created timestamps for the start and end of each detected duration.

for ii = 3:length(logical_array)-1
    if logical_array(ii) == 1 && logical_array(ii+1) == 1 && logical_array(ii+2) == 1 && timepoint(end) ~= "start time"
        count = count + 1;
        if timepoint == ""
            timepoint = "start time";
            epochtime = [epochtime; tab.t(ii)];
        else
            timepoint = [timepoint; "start time"];
            epochtime = [epochtime; tab.t(ii)];
        end
    end
    if logical_array(ii) == 0 && logical_array(ii-1) == 1 && logical_array(ii-2) == 1 && timepoint(end)=="start time" %detects 110, i.e. End Point
       count = 0;
       timepoint = [timepoint; "end time"];
       epochtime = [epochtime; tab.t(ii)];
    end

end
t2.epochtime = epochtime;
t2.timepoint = timepoint;
t2.duration(2:2:length(t2.epochtime)) = t2.epochtime(2:2:end)-t2.epochtime(1:2:end-1);
t2.duration(1:2:length(t2.timepoint)) = NaN;
t2

t2 = 58×3 table
    epochtime     timepoint      duration
    _________    ____________    ________

        13       "start time"      NaN   
        23       "end time"         10   
        25       "start time"      NaN   
        28       "end time"          3   
        37       "start time"      NaN   
        47       "end time"         10   
        58       "start time"      NaN   
        65       "end time"          7   
        72       "start time"      NaN   
        75       "end time"          3   
        93       "start time"      NaN   
        96       "end time"          3   
       101       "start time"      NaN   
       109       "end time"          8   
       111       "start time"      NaN   
       114       "end time"          3   

answer = t2.duration(2:2:length(t2.epochtime))

answer = 29×1

10 3 10 7 3 3 8 3 7 3

[Daniel]

Thank You so much!! It works perfect

Answer 2

[Image Analyst]

If you have the Image Processing Toolbox, you can get the duration of all the pulses in one line of code simple by calling regionprops:

% Measure duration of each region

props = regionprops(mask, 'Area')

Here is the full demo

s = load('tab.mat');
t = s.tab;
x = t.t;
y = t.s;
subplot(2, 1, 1);
plot(x, y, 'b-', 'LineWidth', 2);
grid on;
hold on;
title('Original Data')
% Plot a line at the threshold.
threshold = 3000;
yline(threshold, 'LineWidth', 2, 'Color', 'r')
title('Original Data')
xlabel('t')
ylabel('S')
%-----------------------------------------------------------------------------------
% Get a logical vector for where y exceeds the threshold.
mask = y > threshold;
% Measure duration of each region
props = regionprops(mask, 'Area');
% Get the durations of all regions into a vector (extracting from the structure)
S_3000 = [props.Area]

S_3000 = 1×70

3 2 2 5 1 3 6 1 1 1 1 1 1 5 1 2 1 2 1 2 3 4 4 4 1 1 1 2 2 1

% Show histogram of all the durations
subplot(2, 1, 2);
histogram(S_3000);
grid on;
title('Durations of Pulses')
xlabel('Duration')
ylabel('Count')

Answer 3

[jason mcilvenny]

Another solution:

idx = t.s > 3000;

transitions = diff(idx);

starts = find(transitions == 1);

ends = find(transitions == -1);

durations = ends - starts;

durdates=t.t(starts);

s3000=timetable(durdates, durations);

bar(s3000.durdates, s3000.durations);

datetick