christophm / christophm.github.io Goto Github PK

Hello Christophe,

First of all, I would like to thank you for your book. I first read it last summer and I keep re-opening it whenever I need a refresher.

However, there is one assertion in the section 4.2.4 that I am unsure of: "Like in the linear model, the interpretations always come with the clause that ‘all other features stay the same’ ".

Do we really have to add this clause for linear models too?

In a logistic regression, I understand. Indeed, if Xi and Xj are both increased by one unit each, then the odds is not changed by exp(Bi) + exp(Bj), but by exp(Bi + Bj). So we have to add this clause.

However, in a linear model, an increase in Xi by one unit changes the output Y by Bi units. And this is always the case, even if other variables are not fixed. In other words, if Xi and Xj are both increased by one unit each, then the output Y increases by Bi + Bj. And this happens because of the linearity.

Is there something I am getting wrong here?

Thank you again!