How do I find the time between when the animal_speed exceeded the threshold till when it dipped below the threshold (all instances when this happened)?

The data set I am working with has 17 columns and over 10000 rows. Two particular columns are 'time' and 'animal_speed'. I am also using a Threshold of 0.10m/s.

To find all instances when the 'animal_speed' exceeds the Threshold, I use this:

>> [row, col, v] = find(data_set.animal_speed > 0.10);

which gives me all 'animal_speed' row numbers where the values are greater than 0.10.

Plotting the graph with 'time' and 'animal speed' and including the Threshold line (y=0.10) is not a problem.

However, how do I find the time between when the animal speed exceeded the threshold till when the speed dipped below the threshold (all instances when this happened)?

Thanks!

Answer 1

[Star Strider]

There are likely other solutions, but this seems to work:

animal_speed = rand(1, 1E+4)/2;         % Create Data
time = [0:1E+4]/350;                    % Create Data

Hspeed_log = animal_speed > 0.10;       % Logical Index
Lspeed_log = ~Hspeed_log;
HspeedIdx  = find(Hspeed_log);
LspeedIdx  = find(Lspeed_log);
for k1 = 1:length(HspeedIdx)
    IdxMtx(k1,:) = [HspeedIdx(k1) LspeedIdx(find(LspeedIdx > HspeedIdx(k1), 1, 'first'))];
end
[~,ia] = unique(IdxMtx(:,2));           % Indices Of Unique ‘LspeedIdx’ Values In ‘IdxMtx’
StartStopIdx = IdxMtx(ia,:);            % Start-Stop Index Intervals
StartStopTimes = time(StartStopIdx);    % Start-Stop Times For Speed > 0.10

[SHAUN GEORGE]

Considering I've been using Matlab for barely a few weeks, I still deciphering your answer. =) (It's going to take me a while)

I did however plug your script in and it worked great! It's given me the start and stop times for when the animal_speed was greater than 0.10m/s.

Thank you very much!

[Star Strider]

My pleasure!

The documentation for the find function will explain how it works. The tilde (~) here means logical negation (in the ‘Lspeed_log’ assignment), and ‘don’t return this value to the workspace’ in the unique function call. The logical vectors will be (in this instance) equal to 1 when the animal is moving at greater than 0.1 m/s, and 0 otherwise.

Since you want the time the animal was traveling greater than 0.1 m/s, the first index will always be the beginning of that interval. The ‘IdxMtx’ loop takes that index value and then looks for the next value in the ‘LspeedIdx’ (low-speed index) that is greater.

A slightly different approach would have been to use the diff function on ‘Hspeed_log’, treating the values now as integers rather than logical values (automatic conversion in MATLAB), and looking for the transitions from 0 to 1 equaling 1, and from 1 to 0 equaling -1, then determining those positions in the vector using find. The rest of the code would have been much the same.

My code assumes that the animal has complete (start-stop) activity intervals during the recording, and isn’t moving faster than 0.1 m/s at the end of it. In that event, both times in the last interval will be the same. To prevent that, consider changing the loop to:

for k1 = 1:length(HspeedIdx)
    NextIdx = find(LspeedIdx > HspeedIdx(k1), 1, 'first');
    if isempty(NextIdx)                 % If Moving >0.1 m/s At End, Stop Loop
        fprintf('\n\tIncomplete final interval.\n')
        break
    end
    IdxMtx(k1,:) = [HspeedIdx(k1) LspeedIdx(NextIdx)];
end

This will break out of the loop if the last interval is incomplete, so all the intervals will be valid.

Answer 2

[dpb]

That's certainly an erratic animal!!! :)

Anyway, this is the same question rephrased in slightly different terms just yesterday--

finding-time-intervals-with-corresponding-data

NB: the correction of using ";" for the concatenation operation since your data are also column-oriented. You simply need to turn your speed data into the 0-1 as the switch position the other poster already had; that's simply

speedFlag=speed>=threshold;

where speed is your data set column of interest and threshold the value. Choose '>' or '>=' at your preference of inclusive or no...