I have a model
lm = a ~ b
I would like to include c, d , e that represent interactions terms with b =>
lm = a ~ b + b:c + b:d + b:e.
Is there a rapid way to obtain this result without taping each variable? I have more than 10 variables.
1 Answer
In R formulas, . is a placeholder for all non-response variables present in the data.
coef(lm(mpg ~ wt:(.), data = mtcars))
# (Intercept) wt wt:cyl wt:disp wt:hp wt:drat wt:qsec
# 41.044844014 -16.742131313 0.028771226 0.006485611 -0.005048411 0.447194218 0.423909109
# wt:vs wt:am wt:gear wt:carb
# -0.087023686 0.402891966 -0.142805986 0.156345459
answered May 24, 2022 at 16:35
-
If I want to select specific columns and no all the responses variables ?– ShunreiCommented May 24, 2022 at 19:09
-
2At some point if there are variable you want and variables you don't want, you'll have to let R know which are which. There's probably a way to make it work using the names or even the column numbers of either set (those you want, or those you don't want). Or maybe you can specify a naming pattern (all variables starting with "year_"). If you want more help, ask another question that's more specific, perhaps giving an example of how you'd like to specify the variables you want (or don't want): name, naming pattern, column number, something else? Commented May 24, 2022 at 19:32
a ~ b + b:c + b:d + b:e. You can use parentheses to save a tiny bit of typing,a ~ b + b:(c + d + e). You can use.as a stand-in for all non-response variables in the data,a ~ b:(.). Is that what you're looking for?b:(.)it is then.