WG5 P02 Agenda 2015-08-18
From PHUSE Wiki
Sprint 5 catch-up
- Tues Aug 18th, 2015
- 11:00am EST, 17:00 CET
- 1 hour
- Project 03 update
- Structure of github
- (? demo of working with github desktop?)
- login (github account)
- download/install desktop
- sync phuse project: push, sync basics
- Github structure to be implemented later (this should not prevent our progress)
- R and SAS code can match, but a few details to clarify:
- substitute for analysis flags when analysing SDTM data?
- defining analysis pops and purpose?
- or focus specifically on scripts for analysing ADaM data? (What are our objectives?)
- Output Specifications
- proposal: docx format, draft available in github for CTWP Fig 7.1
- ADaM expertise to justify (provide references) for differences we have in test data (see Fig 7.1 specs)
Testing SAS code - intro to framework for ~70% of tests
- based on PhUSE paper Unit Testing as a Cornerstone of SAS® Application Development
Progress & discussion
- P03 re-structuring could cause problems for people working on components in local repositories
- Gustav plans to attend next P03 meeting, and will raise this topic
- our request is the P03 communicate their plans & schedule in advance
- so we can coordinate our active development around these changes.
- It would also be good for P03 to confirm whether github will preserve change-history of components that move (or is each move in fact a "delete" and "introduce new", which would lose change history)
- We discussed benefits of pursuing the WG5 Box plot challenge via PhUSE newsletter or linkedin group.
- Dante will pursue this with the Communications members of the working group
- All agreed on key benefits, esp. for SAS -- an obscure approach like PROC SHEWHART has several limitations:
- SAS is focused on extending visualization capabilities in other areas, like SGPlot and GTL
- We maximize future potential by finding a robust solution through these techniques
- We also maximize adoption by providing a technique that people want to learn and use.
- Since most Central Tendency data displays are box plots, we need to get the approach right in the code for the first target, Fig. 7.1.
- Peter & Rebeka to peruse the Central Tendency white paper, to identify details for the specification
- Idea is to create an "annotated white paper" that we can use to finalize specs, development scripts, and to support qualification and review
- Jeno & Dante to coordinate an R/SAS script approach
- what scenarios should we focus on for qualification / publication?
- e.g., One R & One SAS script for each white paper target?
- SDTM or ADaM test data?
- What about variations of scripts, e.g., VS box plot from ADaM test data, vs. LB box plot from SDTM data