Hi R user,
I need some help about how to calculate w in a SVR in package e1071.
I have a regression y_i=f(x_i)+e
where f(*x*)=(w,phi(x))+b
then go on with the SVR calculation I know that w*=Sum_i=1^n [(á_i -
á*_i)K(x,x_i) ] where á_i and á*_i are the lagrangian multipliers of the
dual form.
o.k but how I will get it in R?
When I run my model I have information about the model but I don't know how
to use it to get w. Where are the alphas? Is w a vector or a value?
Someone can help me please.
Regards,
Marlene
svm.m1 <- svm(q ~ ., data = train, cost = 100, gamma = 1e-05)
svm.m1
Call:
svm(formula = q ~ ., data = train, cost = 100, gamma = 1e-05)
Parameters:
SVM-Type: eps-regression
SVM-Kernel: radial
cost: 100
gamma: 1e-05
epsilon: 0.1
Number of Support Vectors: 187
[[alternative HTML version deleted]]