The posterior probability of dissolution similarity
View Presentation *David LeBlond, Consultant, CMC Statistics Keywords: The objective of this talk is to consider the problem of dissolution similarity from a Bayesian viewpoint. Aspects of statistical modeling, software tools/code will be presented. Various elements of a Bayesian approach will be contrasted with their traditional counter-parts. It will be argued that 1)defining similarity parametrically is essential, 2) defining the inference space is essential 3) confidence set approaches are inherently conservative, 4) risk based decision making requires estimating the probability of similarity, 5) estimating the probability of similarity demands a Bayesian viewpoint, and 6) the Bayesian approach is conceptually simple and surprisingly easy to implement. Example data sets will illustrate the estimation of the posterior probability of similarity of: 2 batches and 2 processes. Both model independent and dependent approaches will be illustrated. Pros and Cons of the proposed approach will be discussed.
|
Key Dates
-
November 1 - December 17, 2013
Online proposal submission for a session, short course and Town Hall Open -
January 6 - March 11, 2014
Online proposal submission for Roundtables Open -
April 30 - May 28, 2014
Abstract Submission Open -
June 4, 2014
Online Registration Opens -
August 8 - August 22, 2014
Invited Abstract Editing -
August 11, 2014
Short Course materials due from Instructors -
September 1, 2014
Housing Deadline -
September 15, 2014
Cancellation Deadline and Registration Closes @ 11:59 pm EDT -
September 22 - September 24, 2014
Marriott Wardman Park, Washington, DC