Is there an equivalent to R's negative indexing in Matlab?

Question

In R, if we have a vector and a list of indices, we can express the idea that we want "all elements except these indices" using a negative index. In particular, consider the following R code:

data = rnorm(100)
indices = sample(1:length(data), length(data)/2)
training_data = data[indices]
test_data = data[-indices]

After this code, sampled_data contains all the elements in data whose indices are not included in indices. Is there an equivalent to this in matlab?

I tried directly using the same syntax (of course wtih () instead of [], but it just gave the error

Subscript indices must either be real positive integers or logicals.

Answer 1

Matlab does not allow negative indices. What you can do to remove elements is this:

data2 = data;
data2(indices) = [];  % remove selected elements

But when doing machine-learning stuff I prefer to use logical indexing:

istest = randn(length(data), 1) < 0;   % random logicals: 50% 0's and 50% 1's
istrain = ~istest;
% Now operate on data(istest) and data(istrain).

Answer 2

I ended up converting the index array into logicals (rather than generating the logical array directly), because I still wanted the original indexes for other purposes.

indices = datasample(1:length(data), length(data) / 2);
logical = false(length(data) ,1);
logical(indices) = true;
training_data = data(logical)
test_data = data(~logical)

This way of generating logical arrays makes it easier to control the percentage of training vs test examples, at least for me.

However, I still find jez's solution highly educational.