Analysis of nonhomogeneous input data using likelihood ratio test

Abstract

Performing an accurate input analysis in simulation experimentation basically involves selecting the exact probability distributions of random input variables. Frequently in practice these inputs are not constant over time i.e., the underlying distribution may be affected by their time-dependent parameters. In this paper, we propose an approach that can identify whether or not a set of observations follow an identical distribution in a specific period. The model is formulated in a base of likelihood ratio test in the case that input observations come from nonhomogeneous exponentially random variable. Finally, performance comparisons are explored through simulation studies. Keywords -Simulation input data analysis, change point detection, likelihood ratio test, nonhomogeneous Poisson process978-1-4244-4870-8/09/$26.00 ©2009 IEEE

Predicting patient risk of readmission with frailty models in the Department of Veteran Affairs

“…The selected effects are (I) admission source with 2 levels VA hospital and NHCU (Nursing Home Care Unit) as the reference level (II) patient sequence (III) patient LOS (IV) marriage status with 3 levels 'married', 'previously married', and 'never married' as the reference category (V) user enrollment status with 2 levels 'YES' and 'NO' as the reference class. We also examined Schwarz's criterion (also known as BIC) to the dataset and found that the model, which minimizes this criterion, again contain the above features [18,19].…”