This is a Python implementation of the Sison-Glaz (1995) method for constructing simultaneous confidence intervals for multinomial proportions. This is a translation of the source code used to implement the "MultinomialCI" R package (source code: https://github.com/cran/MultinomialCI, originally prepared by Dr. Pablo Villacorta), which itself was an R-based implementation of a SAS package coded by May and Johnson (2000). An example usage is provided in the definition of the "main" function at the bottom of this script.
Sison, C. & Glaz, J. (1995). Simultaneous Confidence Intervals and Sample Size Determination for Multinomial Proportions, Journal of the American Statistical Association, 90:429, 366-369, DOI: 10.1080/01621459.1995.10476521
May, W., & Johnson, W. (2000). Constructing two-sided simultaneous confidence intervals for multinomial proportions for small counts in a large number of cells. Journal of Statistical Software, 5:6, 1-24. doi:http://dx.doi.org/10.18637/jss.v005.i06 [Online: see https://www.jstatsoft.org/article/view/v005i06]
R package for MultinomialCI, by P. Villacorta: https://cran.r-project.org/package=MultinomialCI
Brian Canada, PhD (bcanada@uscb.edu)
Date of original commit: 31 July 2019