makeLearnQ.piecewise: Builds a Parametric Model Based on Sample Size

In achambaz/tsml.cara.rct: Targeted Sequential Minimum Loss CARA RCT Design and Inference

Description

Builds a Parametric Working Model Based on Sample Size

Usage

makeLearnQ.piecewise()

Details

This functions builds a sample-size-dependent parametric working model. Two fine-tune parameters are determined based on sample size n: deg=\min(\lceil n/100\rceil, 6) and nlev=\min(\lceil n/100\rceil, 10). If nlev equals one, then the model is given by the formula

Y~I(A=0)*V*poly(U)+I(A=1)*V*poly(U)

where poly(U) consists of deg orthogonal polynoms of degrees 1 to deg. If nlev is larger than or equal to two, then the model is given by the formula

Y~I(A=0)*V*(poly(U)+∑_k step_k (U))+I(A=1)*V*(poly(U)+∑_k step_k (U))

where poly(U) consists of deg orthogonal polynoms of degrees 1 to deg and step_k (U) consits of nlev ortogonal polynoms of degrees 1 to deg in U times the indicator function of the subset of [0,1] of the form \{x:\frac{k}{nlev}≤ x <\frac{k}{nlev}\}.

Value

Returns a formula describing the parametric model.

Author(s)

Antoine Chambaz [aut, cre]

achambaz/tsml.cara.rct documentation built on May 10, 2019, 5:10 a.m.