Develop uncertainty estimates using Monte-Carlo simulation, bootstrap-resampling

Dear Matlab users,

I have recently engaged with an issue related to calculating uncertainty for some defined parameters using 10 years of data (each year separately).

As per the reference model (or truth dataset), it uses all this 10 years data and calculate required statistical parameters, but for the current uncertainty analysis, it should be taken either one year or two years or more years of data to calculate the related parameters and develop the uncertainty population (see the attached figure). It’s better to develop more than 1000 datasets as you all aware in terms of Monte-Carlo simulations and bootstrap resampling.

Can anyone help me to solve this issue with selecting datasets? The main consideration here is to take yearly datasets as slots.

Hope you understand my question.

Thank you!

Answer 1

[Jeff Miller]

You seem to be asking how to select a random subset of the years for each iteration of the simulation. If that is right, then you can probably just use something like this:

for isim=1:nsims
    yrs = randperm(10);
    % Now use the years indicated by yrs(1) and yrs(2), for example,
    % if you want the simulation to use two random years from the original 10-year set.
end