biClassify (version 1.3)

SelectParams: Generates parameters.

Description

Generates parameters to be used in sparse kernel optimal scoring.

Usage

SelectParams(TrainData, TrainCat, Sigma = NULL, Gamma = NULL,
  Epsilon = 1e-05)

Arguments

TrainData

(n x p) Matrix of training data with numeric features. Cannot have missing values.

TrainCat

(n x 1) Vector of class membership. Values must be either 1 or 2.

Sigma

Scalar Gaussian kernel parameter. Default set to NULL and is automatically generated if user-specified value not provided. Must be > 0. User-specified parameters must satisfy hierarchical ordering.

Gamma

Scalar ridge parameter used in kernel optimal scoring. Default set to NULL and is automatically generated if user-specified value not provided. Must be > 0. User-specified parameters must satisfy hierarchical ordering.

Epsilon

Numerical stability constant with default value 1e-05. Must be > 0 and is typically chosen to be small.

Value

A list of

Sigma

Gaussian kernel parameter.

Gamma

Ridge Parameter.

Lambda

Sparsity parameter.

Details

Generates the gaussian kernel, ridge, and sparsity parameters for use in sparse kernel optimal scoring using the methods presented in [Lapanowski and Gaynanova, preprint]. The Gaussian kernel parameter is generated using five-fold cross-validation of the misclassification error rate aross the .05, .1, .2, .3, .5 quantiles of squared-distances between groups. The ridge parameter is generated using a stabilization technique developed in Lapanowski and Gaynanova (2019). The sparsity parameter is generated by five-fold cross-validation over a logarithmic grid of 20 values in an automatically-generated interval.

References

Lancewicki, Tomer. "Regularization of the kernel matrix via covariance matrix shrinkage estimation." arXiv preprint arXiv:1707.06156 (2017).

Lapanowski, Alexander F., and Gaynanova, Irina. ``Sparse feature selection in kernel discriminant analysis via optimal scoring'', Artificial Intelligence and Statistics, 2019.

Examples

# NOT RUN {
Sigma <- 1.325386  #Set parameter values equal to result of SelectParam.
Gamma <- 0.07531579 #Speeds up example

TrainData <- KOS_Data$TrainData
TrainCat <- KOS_Data$TrainCat

SelectParams(TrainData = TrainData , 
             TrainCat = TrainCat, 
             Sigma = Sigma, 
             Gamma = Gamma)
# }