How can I get the list of non-zero features from svmpath at any given
lambda? All I get is following information and information about what
features were selected. In Iris example, we have 4 features and 60 cases. In
my own example which is 200cases by 300 features, I can't figure out how to
print the list of non-zero (beta) or features from svmpath.
Any help? Thanks.
----------------
> solution <- svmpath(as.matrix(iris[1:60,1:4]), buffer,
type="class",
trace=FALSE)
> names(solution)
[1] "alpha" "alpha0" "lambda"
"alpha00"
"Error" "SumEps"
[7] "Size.Elbow" "Elbow" "Moveto"
"Movefrom"
"Obs.step" "Step"
[13] "kernel" "param.kernel" "x"
"y" "call"
> solution$alpha0
[1] 130.785000 108.685000 106.210000 86.130000 85.485000 71.176837
68.197549 67.660686
[9] 60.847899 58.442887 53.721098 53.544146 44.610000 44.525517
44.415343 39.709048
[17] 38.832267 36.940401 33.800161 28.115484 25.859545 25.734545
24.818367 19.666905
[25] 18.833000 17.768614 17.219663 16.737715 15.503339 14.717148
8.586288> solution$Obs.step
[1] 27 53 27 53 32 51 32 51 26 55 6 55 57 6 44 57 59 44 19 44 59 52 19 44
21 52 56 21 52 42 26
[32] 56 45 52 54 42
[[alternative HTML version deleted]]
