Optimal allocation with known covariates into two treatments under generalized linear model through Hybrid VNS algorithm
Article Type
Research Article
Publication Title
Communications in Statistics: Simulation and Computation
Abstract
The problem of optimal allocation of experimental units with known covariates into several treatment groups under linear ANCOVA model has already been discussed by the same authors and the optimal allocation design has been derived through an efficient algorithm named as hybrid variable neighborhood search (VNS) algorithm. In this work, we have addressed the same issue with regard to D- and A-optimality under the generalized linear model (GLM) set-up, assuming the response variable to be count or binary. The issue of mean covariate balance is also addressed through a constrained optimal approach in which a compromise between balance and optimality is maintained. The performance of the optimal and constrained optimal allocation design obtained through the hybrid VNS algorithm, under various GLM frameworks, has been investigated through simulation studies and real-life example.
DOI
https://10.1080/03610918.2023.2250585
Publication Date
1-1-2023
Recommended Citation
Hore, Samrat; Dewanji, Anup; and Chatterjee, Aditya, "Optimal allocation with known covariates into two treatments under generalized linear model through Hybrid VNS algorithm" (2023). Journal Articles. 3934.
https://digitalcommons.isical.ac.in/journal-articles/3934
Comments
Open Access, Green