# Wiki of the LARS tag

This is the current Wiki of the tag:

LARS is an extension of the LASSO, which constrains regression coefficients to no more than a possible absolute sum. The LARS algorithm can be understood as recalculating the regression model step by step while slowly relaxing the LASSO constraint. The result of this is analogous to a forward stepwise selection algorithm, but is valid, whereas forward stepwise selection is not.

I have a minor quibble and a major issue:

1. "Recalculating" the regression model -- is the verb appropriate here, or should we say "reestimate" or similar? (I don't know, but it looks suspicious.)
2. "The result of this is analogous to a forward stepwise selection algorithm, but is valid, whereas forward stepwise selection is not." How is that? In what sense is it analogous? In what sense is it valid? The statement seems to broad to be informative or useful.

I hope someone who knows more about LARS will edit the Wiki accordingly.

• The wiki was written by @gung (stats.stackexchange.com/posts/47985/revisions), he will hopefully comment here. – amoeba May 9 '17 at 19:57
• If you want to rephrase it, that's fine with me. Reestimating seems better than recalculating. The sense in which it might be considered analogous to forward selection is probably hard to explain in an excerpt; if you can think of a better phrasing, that's fine, or we could delete that comment. – gung - Reinstate Monica May 11 '17 at 11:36
• @gung, I assume you know more about LARS than I do; I only have superficial knowledge of it. I think it is better if you edit it. I would just encourage to either include accurate and concrete statements or nothing (delete the lacking ones) rather than keeping the somewhat inaccurate and potentially misleading information (as per my question, i.e. the OP). – Richard Hardy May 11 '17 at 11:52

I edited the wiki for . It now reads:

LARS is an extension of the LASSO, which constrains regression coefficients to no more than a possible absolute sum. The LARS algorithm can be understood as reestimating the regression model step by step while slowly relaxing the LASSO constraint.

The result of this is analogous to a forward selection selection algorithm in that the first variable included would be the one most strongly associated with the response, and as the constraint is relaxed, additional variables would be included in descending order of their strength of association.

Specifically,

1. I switched "reestimating" for "recalculating".