performanceBLB: Performance comparison of severals estimation algorithms

In jgbabativam/BiplotML: Biplots Estimation with Algorithms ML

Performance comparison of severals estimation algorithms

Description

This function computes the estimates of A and B matrix with severals algorithms.

Usage

performanceBLB(xi, k = 2, L = 0, method = NULL, maxit = NULL)

Arguments

xi	Binary matrix.
k	Dimensions number. By default k = 2.
L	Penalization parameter. By default L = 0.
method	use value 1 for algorithms without gradient, 2 with gradient, 3 quasi-newton methods or 4 for all methods. By default method = 2.
maxit	The maximum number of iterations. Defaults to 100 for the gradient methods, and 500 without gradient.

Details

This function compare the process time and convergence of different algorithms without gradient, with gradient or quasi-newton method for estimating the parameters in a Binary Logistic Biplot

Value

data frame with method, time of process, convergence and number of evaluations

Author(s)

Giovany Babativa <[e-mail]>

References

John C. Nash (2011). Unifying Optimization Algorithms to Aid Software System Users:optimx for R. Journal of Statistical Software. 43(9). 1–14.

John C. Nash (2014). On Best Practice Optimization Methods in R. Journal of Statistical Software. 60(2). 1–14.

Vicente-Villardon, J.L. and Galindo, M. Purificacion (2006), Multiple Correspondence Analysis and related Methods. Chapter: Logistic Biplots. Chapman-Hall

See Also

gradientDesc

Examples

data('Methylation')
set.seed(123456)
########### Gradient Methods
performanceBLB(xi = Methylation)
performanceBLB(xi = Methylation, maxit = 150)

########### Without Gradient Methods
performanceBLB(xi = Methylation, method = 1)
performanceBLB(xi = Methylation, method = 1, maxit = 100)

########### Quasi-Newton Methods
performanceBLB(xi = Methylation, method = 3)
performanceBLB(xi = Methylation, method = 3, maxit = 100)

########### All methods
performanceBLB(x = Methylation, method = 4)

jgbabativam/BiplotML documentation built on July 31, 2022, 11:10 a.m.