I have been working with quantreg::rqss() function for non-parametric quantile regression in R (documentation is here).
And I have a question of what is happening behind the scene with argument lambda. I understand it is a smoothening term , but I wonder if someone can explain in plain language what kind approach it takes to "smooth" the curve. For example, is it taking the nearest neighboring points like lowess does? or it is something different?
I do not have background in statistics, so reading a suggested paper does not make it clear for me.
Thank you a lot in advance!
