Package src :: Package stats306b :: Package lecture10 :: Module fda

Module fda

source code

Flexible discriminant analysis (FDA)

Functions

mlm(X, Y, l=0.0, Omega=0.0)
Multivariate linear regression model of X on Y with optional ridge penalty on each column of beta.

source code

fda(X, labels, method=<function mlm at 0xab9556c>, **methodargs)
Fit FDA model with a given regression method (taking optional keyword arguments)

source code

main()
Demonstration to show the FDA with the standard regression model gives LDA

source code

Function Details

Multivariate linear regression model of X on Y with optional ridge penalty on each column of beta.

Note: penalty is enforced identically to each column of beta

Fit FDA model with a given regression method (taking optional keyword arguments)

"method" should return a function that takes an X value and gives an estimator: Yhat, where Yhat is a vector of indicators for the given labels.