perform join without using dataset or table

Hi everyone,

I am trying to add weights to a set of observations based on their year. So, I have 150 observations across 30 unique years. I have one weight per year. I would like to broadcast the weights to the 150 observations linking based on year. I understand that this can be done using database join functionality but I am wondering if there is a way to do this just using matrices.

To restate: My weights matrix is 30x2 (unique years x [years weights]). My data matrix is 150x2 (num observations x [year blank-waiting-for-weight].

Computational efficiency is critical here so that's why I'm trying to avoid conversions if possible.

Thanks for your thoughts, Mike

Answer 1

[dpb]

[~,lib]=ismember(dat(:,1),w(:,1));  % location of year in data
dat=[dat w(lib,2)];                 % concatenate the weights in those rows

[Mike Alonzo]

Perfect. Thanks. Forgot about the second output of ismember.