*I have an overdetermined system Ax = b + ε.

Given A and b, what is the recommended way to find x which minimizes the L1 norm of ε ?

*I have an overdetermined system Ax = b + ε.

Given A and b, what is the recommended way to find x which minimizes the L1 norm of ε ?

I borrowed the text “Robust Statistical Methods with R” from Michigan Tech.

The bottom of page 124 gives an example of LAD estimation using rq() from the *quantreg* package.

Can this function be used to do LAD estimation for the above stated problem?

**A** ∈ ℝmn m>n … **b** & **ε** ∈ ℝm … **x** ∈ ℝn

I’ve only used R for some simple transformations and stats on large data sets.

You might be better off on the R-Help list and asking your question there.

Good luck!

I vaguely recall from a couple of years ago that there were 1 or 2 folks here on CD who said they used R extensively in their work. Can’t remember the names though…

You might be better off on the R-Help list and asking your question there.

I submitted the question to that mailing list and it sat in the queue for 10 days awaiting moderation before finally being rejected as “spam”. Go figure.