2
$\begingroup$

I couldn't find any topic with this discussion. Not sure if this is just too obvious and people don't even discuss, but I'll take my chances... does linear regression have four or ten assumptions? I would guess that the ten assumptions somehow mean the same and you actually don't need all of them. But I'm not sure yet, I couldn't prove this.

I've been discussing linear regression for a long time, because I can't understand the relation with the assumptions. Lots of Machine Learning tutorials don't even care about the assumptions. Isn't it important? Why there are so many tutorials that don't analyze linear regression assumptions?

Improve this question
$\endgroup$
1
  • 3
    $\begingroup$ The ten assumptions (actually 8) are highly overrestrictive. If they were really required, very few regression models would ever be used. It's more important to understand the assumptions, to know how to check them in any instance, and to evaluate the possible effects of departures from them. Counting them has little interest. $\endgroup$
    – whuber ♦
    Aug 24, 2021 at 22:05

0

Reset to default

Browse other questions tagged or ask your own question.