backward.selection function

Backward selection function for the second screening step

Backward elimination algorithm

backward.selection(data, pts, lambda, mu, alpha_L = 0.25)

Arguments

• data: A n by p dataset matrix
• pts: A numeric vector, which includes all candidate change points obtained from the first step
• lambda: Tuning parameter for sparse component
• mu: Tuning parameter for low-rank component
• alpha_L: Constraint space for low rank component, default is 0.25

Returns

A list object, containing

• L.n: Value of objective function
• L.n.current: Current value of objective function