Choose Language Hide Translation Bar
Chrismtoy
Staff (Retired)

A Major JMP® Scripting Language Application for Decision Tree Analysis

 A Major JMP® Scripting Language Application for Decision Tree Analysis

 

David Burnham, Consultant, Pega Analytics
Martin Owen, Quality by Design Innovation Leader; Gill Turner, Quality by Design Scientist, GlaxoSmithKline

We all have to make decisions. And usually we have to make them in the face of uncertainty. If we invest in R&D activities, we can never be certain of the outcome. Will the work we plan to do reduce risk and/or increase benefits? If so, when is the best time to invest — now, or when we have a greater reason to believe the product will launch successfully? To help us model and visualize the outcomes of several possible scenarios, we have developed a JMP Scripting Language (JSL) application to build and analyze decision trees. This is a major piece of programming involving approximately 15,000 lines of object-oriented JSL code. This new application guides the user through the process of structuring decisions and analyzing the sensitivity of the factors that determine the magnitude and likelihood of different possible outcomes. And that's just for one decision; in reality we deal with portfolios of decisions. In the pharmaceutical industry, for every 100 potential drugs that enter product development less than 10 will become marketed drugs. The great uncertainty lies in not knowing which drugs will make it and which will undergo attrition. In drug development, our role is not to control the attrition, but to manage the consequences. In this presentation we will: 1) explain why we have implemented the workflow to enable open and transparent decision making within project teams; 2) describe the design principles and the construction underpinning this new JMP application.

Article Tags
Contributors