Question

Please refer to pp. 33-34 Example 1.18 The False-Positive Puzzle http://www.cs.utexas.edu/~jason777/School/introduction%20to%20probability%20(bertsekas,%202nd,%202008).pdf

I too arrived at 0.0187 there. However, simulating a model of this in both R and matlab shows a converergence to 0.0197~0.02 instead of 0.0187. Would someone mind looking at the following matlab code to see if I've modelled it correctly? This is for a 10,000 sample population (also tried with other samples, same result) -

Answer 1

function out = carloexample(trials)

for j = 1:trials
positives=0;
population=[zeros(9991,1); ones(9,1)];
for i = 1:length(population)
r=rand(1);
if population(i) == 0
if r >=0.05
elseif r<0.05
positives=positives+1;
end
elseif population(i) == 1
if r>=0.05
positives=positives+1;
elseif r<0.05
end
end

end

total(j)=positives;
end
out=1/mean(total);
end