API Documentation
Documentation for ExperimentalDesign.jl
's API.
Contents
Index
ExperimentalDesign.AbstractRandomDesign
ExperimentalDesign.AbstractResponseSurfaceDesign
ExperimentalDesign.IterableFullFactorial
ExperimentalDesign.RandomDesign
ExperimentalDesign.a_criterion
ExperimentalDesign.e_criterion
ExperimentalDesign.g_criterion
ExperimentalDesign.rd_criterion
ExperimentalDesign.t_criterion
API
ExperimentalDesign.AbstractRandomDesign
— Typeabstract type AbstractRandomDesign <: AbstractDesign
ExperimentalDesign.AbstractResponseSurfaceDesign
— Typeabstract type AbstractResponseSurfaceDesign <: AbstractDesign
ExperimentalDesign.IterableFullFactorial
— Typestruct IterableFullFactorial <: AbstractFactorialDesign
matrix::DataFrame
factors::NamedTuple
formula::FormulaTerm
ExperimentalDesign.RandomDesign
— Typestruct RandomDesign <: ExperimentalDesign.AbstractRandomDesign
matrix::DataFrame
factors::NamedTuple
formula::FormulaTerm
ExperimentalDesign.a_criterion
— Methoda_criterion(model_matrix; tolerance) -> Any
Criterion of A-optimality which seeks minimum of $trace((X^T · X)^{-1})$. This criterion results in minimizing the average variance of the estimates of the regression coefficients.
Criterion metric is $\frac{p}{trace((X^T · X)^{-1}) · N}$.
ExperimentalDesign.e_criterion
— Methode_criterion(model_matrix; tolerance) -> Any
Criterion of E-optimality maximizes the minimum eigenvalue of the information matrix ($X^T · X$).
Minimization metric is $\frac{\min λ(X^T · X)}{N}$
ExperimentalDesign.g_criterion
— Methodg_criterion(model_matrix; tolerance) -> Any
Criterion of G-optimality which seeks minimize the maximum entry in the diagonal of $X·(X^T · X)^{-1}·X^T$. This criterion results in minimizing the maximum variance of the predicted values.
Minimization metric is $\frac{N}{\max diag(H)}$ where $H = X·(X^T · X)^{-1}·X^T$.
ExperimentalDesign.rd_criterion
— Methodrd_criterion(model_matrix; tolerance) -> Any
Criterion of rotation distance optimality. Experimental.
ExperimentalDesign.t_criterion
— Methodt_criterion(model_matrix; tolerance) -> Any
Criterion of T-optimality. This criterion maximize $trace(X^T · X)$.
Minimization metric is $\frac{trace(X^T · X)}{N · p}$.