Reducing computation time for large data set

Pavlo Ivanusa

2017 6 9

0 个回答

5 次查看（30 天）

0 个投票

I'm looping through a large data set (1843200 data points and a total data size of ~30MB). Basically, I am using a for loop to step through each data point, and then create a temporary vector that is a subset of the original data vector. I then use the average and RMS functions to find a value which I add to a third array. The code currently takes about 60 seconds to run, which is not awful by itself, but it does present problems when I have to run this code of several thousand data sets. Below is the piece of code in question. Is there any way I can speed up the process, e.g. looping through differently or appending to arrays, etc?

function [rmsVector,timeVector] = rms_calc(colVector, timeSpan, sampleFreq)
%NOTE: timeSpan must be in seconds and sampleFreq must be in samples/sec
vecLen = length(colVector); %vector length
timeLen = sampleFreq*timeSpan; %converts time to indices
actualLen = vecLen-timeLen; %actual length of output data.
divRem = rem(vecLen,timeLen); %find the remainder
if divRem ~= 0
    fprintf('The time of the data is not evenly divisible by the chosen time span. This may cause inaccuracies in the RMS calculations')
end
rmsArray = zeros(actualLen,1);
timeArray = zeros(actualLen,1);
for i = 1:actualLen
    colStart = i;
    colEnd = colStart+timeLen-1;
    tempArray = colVector(colStart:colEnd);
    meanVal = mean(tempArray);
    rmsArray(i) = rms(tempArray-meanVal);
    timeArray(i) = i/sampleFreq+timeSpan;
    if rem(i,20000.0)==0
        fprintf('%3.0f%%\n',i/vecLen*100); %I'm just using this to keep track of the code and which iteration
    end
end
rmsVector = rmsArray;
timeVector = timeArray;