# Calculate the Median of the results from 100 Simulations

1 view (last 30 days)
CMatlabWold on 13 Oct 2021
Commented: CMatlabWold on 14 Oct 2021
Hi. I have a code where I am running a Random Forest regression. I am running it 100 times. However, I am having difficulty calculating the median of the 100 trials.
The result I am looking for is located in the variable designated "impOOB".
For each run, there should be values in impOOB variable for 5 columns. For instance:
0.427417559041683 0.00894308188405568 0.141297948087486 0.222153283589539 0.200188127397237
For 100 runs of column 1, I need the median. The same for column 2, and so forth.
My code is as follows:
n = 100;
result = zeros(n,5);
for k=1:n
t = templateTree('NumVariablesToSample','all',...
'PredictorSelection','interaction-curvature','Surrogate','on');
Mdl = fitrensemble(X,Y,'Method','Bag','NumLearningCycles',200, ...
'Learners',t);
yHat = oobPredict(Mdl);
R2 = corr(Mdl.Y,yHat)^2
impOOB = oobPermutedPredictorImportance(Mdl);
impOOB(impOOB<0) = 0;
impOOB = impOOB./sum(impOOB)
result(k) =
end
I'll attach the files as well. I appreciate very much any help with this.

Matt J on 14 Oct 2021
Edited: Matt J on 14 Oct 2021
impOOB=rand(100,5)
impOOB = 100×5
0.7604 0.5152 0.7196 0.2418 0.5420 0.8386 0.3787 0.4701 0.4692 0.7572 0.9929 0.0561 0.2087 0.1176 0.9434 0.7796 0.2491 0.1337 0.1499 0.9048 0.3695 0.1500 0.6826 0.4575 0.4751 0.6069 0.1207 0.8111 0.5832 0.3273 0.9885 0.2647 0.1840 0.9606 0.0610 0.1243 0.3249 0.5171 0.1649 0.9400 0.7085 0.7869 0.5282 0.5472 0.4634 0.7656 0.4034 0.7932 0.8618 0.5136
median(impOOB,1)
ans = 1×5
0.4813 0.4997 0.4696 0.4513 0.5518
##### 3 CommentsShow 1 older commentHide 1 older comment
Matt J on 14 Oct 2021
n = 100;
result = zeros(n,5);
for k=1:n
t = templateTree('NumVariablesToSample','all',...
'PredictorSelection','interaction-curvature','Surrogate','on');
Mdl = fitrensemble(X,Y,'Method','Bag','NumLearningCycles',200, ...
'Learners',t);
yHat = oobPredict(Mdl);
R2 = corr(Mdl.Y,yHat)^2
impOOB = oobPermutedPredictorImportance(Mdl);
impOOB(impOOB<0) = 0;
result(k,:) = impOOB./sum(impOOB);
end
median(result,1)
CMatlabWold on 14 Oct 2021
It works. Thank you very much

### Categories

Find more on MATLAB in Help Center and File Exchange

R2021a

### Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!