MADlib Roadmap

Jump to bottom Edit New page

cwelton edited this page May 9, 2012 · 8 revisions

MADlib Development timeline

MADlib is currently on a quarterly release cadence with monthly development sprints.

MADlib v0.4

March 1st: Call for features
- Random Forest - Studies indicate significant accuracy improvement over the existing C4.5 decision tree implementation.
- K-means - Refactoring for performance and extended distance metric support.
- Variable Selection - Some research still needed to finalize precise methods.
- Statistical Significance tests - extremely important additions
- Viterbi feature extraction - in progress from previous release.
March 15th: Design Specifications finalized for MADlib v0.4
April 12th: Status Checkpoint
- Final evaluation for inclusion of proposed features
- Any proposed features not sufficiently far along development moved to next release
- Test specifications complete
May 10th: Code Complete
May 31st: Testing certification complete
- Any new feature not satisfying quality bar deferred to v0.5
June 7th: Ship date MADlib v0.4

MADlib v0.5

May 31st: Call for features
- Any features that did not make the cut for MADlib v0.4
- Variable Selection - Some research still needed to finalize precise methods.
- CRF methods to support text extraction work
- Multivariate Logistic Regression
- Marginal effects for ols, logistic, and multivariate logistic
- TBD
June 21st: Design Specifications finalized for new content
July 19th: Status Checkpoint
Aug 9th: Code Complete
Aug 30th: Testing certification complete
Sep 6th: Ship date MADlib v0.5

MADlib v1.0

Dec 6th: Ship date MADlib v1.0