Timeline for answer to difference between feature interactions and confounding variables by Peter

4 events

when toggle format	what		by	license	comment
Dec 31, 2019 at 7:37	comment	added	The Great		w.r.t point 4, by interaction what I was interested to know was that, none of the algorithms has this as their property. Am I right? I mean it has to be done by me to select 2/3/4 features and decides to multiply/add/subtract etc. This is called creating interaction variables. Am I right? There is no algorithm that does this by itself. Or is there any ML model that considers interactions between variables by default?
Dec 31, 2019 at 7:34	comment	added	The Great		Regarding point 5, I meant this approach to adjust confounders. You can refer this link sphweb.bumc.bu.edu/otlt/MPH-Modules/BS/… My question here was that, this is not just the only way to adjust confounders and we can do it using multiple regression as well. Am I right?
Dec 31, 2019 at 0:23	comment	added	The Great		Hi, thanks for the response. Upvoted. regarding point 2, yes, I am aware and understand what does confounder mean. I have given an example of `Age`. I read in literature that few ways to control confounders. One is through `startified analysis` and other is through `multivariate regression`. In my case, am using `multivariate regression`, so is it only about putting all variables (which will include confounders as well) in a regression model which is known as confounding adjustment through regression?
Dec 30, 2019 at 17:46	history	answered	Peter	CC BY-SA 4.0