Sharing Standard Statistical Scripts: Simple Slides A Journey in Progress September 18 2012 Michael Carniello [email protected] Subgroup 3 of.

Sharing Standard Statistical Scripts: Simple Slides

A Journey in ProgressSeptember 18 2012Michael Carniello

[email protected] 3 of PhUSE Working Group 5

The Question

• What platforms are available to use a medium for developing, publishing and maintaining Standard Statistical Scripts – with world-wide developers being in industry, academia and government?

• In FDA/PhUSE Meeting in early 2012, Subgroup 3 of Working Group 5 took on this question

Well, It’s a Start

• Ian Fleming, Hanming Tu and Mike Carniello spent some time working as a sub-subgroup and looked hard at Github

• They also looked at GoogleDocs, as Github proved to be fairly intense

• Right now (September 18 2012), Mike’s opinion is that GoogleDocs provides a reasonable starting place for such a platform

The Rest of This Slide Deck

• Some information about Github, how it looks and works, and why it was rejected

• Some information about GoogleDocs, how it looks and works, and why it seems like a reasonable first approach

Github

• Used to develop Facebook, Twitter, etc.• From Wikipedia

Github: It’s Complicated

Git/Bash shell for Windows

Github in a Browser

Github in a Browser (Deeper)

Editing a Github file in a Browser

Looks Good; What’s the Problem?

• In testers’ opinions, too much for classic statistical programming users– Too complicated an interface– Too much overhead for simple development– Too much training and education needed

• Fair enough; Github designed for classic programming languages like C and Java (not for things like R and SAS)

GoogleDocs

• Designed for collaboration on a variety of documents• Bar is low with respect to learning and using• Advantage – not horriby sophisticated• Disadvantage – not horribly sophisticated• From Wikipedia:

A Directory View in a Browser

Editing a File in a Browser(with Revision History)

More Editing in a Browser

First Real Target:Efficacy by Site Figure

• R code (snip) from CTSpedia

Efficacy by Site Figure

• SAS code (snip) on GoogleDocs

Efficacy by Site: R Output

Efficacy by Site: SAS Output

The End ???

• Not hardly!• Mike’s suggested next steps:– 1. Pick a finite list of scripts to develop– 2. Assign primary authors and set

milestones/deadlines for GoogleDoc deliveries– 3. After last development deadline, post to

PhUSEWiki