Reputation: 23
I have a quadratic model that I am trying to fit with mgcv::gam
, and I need some help in coding it as an r formula
:
My attempt so far:
model <- mgcv::gam(W ~ 1 + z + I(0.5*z^2))
Do you think this is the correct syntax for the formula? That is, does the formula
match the equation I am trying to use? I am not exactly sure whether the 0.5
is placed correctly.
Upvotes: 0
Views: 79
Reputation: 1320
The 1
in the formula will be interpreted as an unknown intercept to be estimated, not as a constant of 1. I think W ~ -1 + offset(rep(1, length(z))) + z + I(0.5*z^2)
will do what you want. The -1
specifies that there is no intercept and offset(rep(1, length(z)))
serves as the 1 in your equation. This works for stats::glm
, so it should also work for mcgv::gam
according to the help page, but I'm not familiar with mcgv
so there might be quirks I'm not aware of.
Upvotes: 3