sc1-svm-example-package

This is a small package implementing soft and hard margin SVMs with abitrary kernels using the quadratic programming library quadprog

Installation

library(devtools)
install_github("alessio-b-zak/sv1-svm-example-package")
library(alessiosvm)

Usage

This package can be used by invoking the following function:

svm(X=X, 
    classes=classes, 
    C=C, 
    margin_type=margin_type, 
    kernel_function=kernel_function, 
    feature_map=feature_map)

where:

• X is a data matrix with observations on rows
• classes are the labels associated with each observation of the data matrix (either 1 or -1)
• C is the numeric cost associated with the soft margin classifier (not needed if `margin_type == "hard")
• margin_type is either hard or soft
• kernel_function is the kernel function to build the kernel matrix (not needed if `margin_type == "hard")
• feature_map is the map corresponding the the induced feature space of the kernel function (not needed if `margin_type == "hard")

svm() returns a model object which contains $prediction_function which classifies a new data point and $params which contains the w and b parameters associated with predition_function