Closed Bug 1400184 Opened 8 years ago Closed 8 years ago

TAAR: Compare ensemble methods

Categories

(Data Platform and Tools :: General, enhancement)

Product:
Data Platform and Tools
Component:
General
Type:
enhancement
Priority:
Not set
Severity:
normal

Tracking

(Not tracked)

Status:
RESOLVED FIXED

People

(Reporter: fhartmann, Assigned: fhartmann, Mentored)

References

Details

Some research should be done to find out which Machine Learning ensemble algorithms are well suited for recommender systems, and could work well for TAAR. The results should be documented and later presented, to decide which of these algorithms we want to use.
Blocks: 1400186
The document that summarizes my research can be accessed using a Mozilla account [1]. To find an appropriate ensemble algorithm, I went down two paths: First of all, I looked at very general ensemble methods and tried to figure out how well they would work on our recommender system. Then, I looked at specific literature about ensembles for recommender systems. Most of that research came out of the Netflix Prize competition. Moving forward, it looks like a linear stacking or blending ensemble is most appropriate for our use case. [1] https://docs.google.com/a/mozilla.com/document/d/1EAjrPn1FP_em7bLP24V9Bys-HHv5U-VauuLhzOXU4OE/edit?usp=sharing
Status: NEW → RESOLVED
Closed: 8 years ago
Resolution: --- → FIXED
Moved to new component, per bug 1425844.
Component: General → Add-on Recommender
Component: Add-on Recommender → General
You need to log in before you can comment on or make changes to this bug.