MLEw2p_abrem: Weibull 2-parameter MLE calculation using differentiation of...

In debias: Functions implementing MLE,its bias reduction for small samples and related issues

Description

MLEw2p_abrem fits a set of data consisting of failures, or alternatively failures and suspensions, to the 2-parameter Weibull distribution.

Usage

 MLEw2p_abrem(x, s=NULL, MRRfit=NULL, limit=1.0e-6, listout=FALSE)

Arguments

x	A vector of failure data.
s	An optional vector of suspension data.
MRRfit	An optional vector such as produced by MRRw2pxy having parameter order [1] Eta, [2] Beta. If not provided, this function will calculate a suitable estimate of Beta to initiate the optimization.
limit	A limiting value for the optimization. Default 1.0e-6 should serve most purposes.
listout	A boolean flag for permitting an alternate output as a list containing the primary result vector and a dataframe providing details of the optimization search.

Details

This function will sort all input data while preserving failure and suspension information in paired vectors. The code is expository as it implements the root identification of the derivative of the likelihood function with respect to Beta, then given the optimal Beta calculate Eta from as the root of the derivative of the likelihood function with respect to Eta. The optimization algorithm employed is a discrete Newton, or secant, method as demonstrated in a FORTRAN program published by Tao Pang.

Value

A vector containing results in the following order: Eta (scale), Beta (shape), Log-Likelihood). This vector is returned by default and also as list item [[1]] depending on the listout argument. The second optional list item is a dataframe providing details of the optimization search.

References

Dr. Robert B. Abernethy, (2008) "The New Weibull Handbook, Fifth Edition" Tao Pang,(1997) "An Introduction to Computational Physics"

Examples

failures<-c(90,96,30,49,82)
suspensions<-c(100,45,10)
fit_result<-MLEw2p_abrem(failures,suspensions)

debias documentation built on May 2, 2019, 4:49 p.m.