User Metrics

Group:	Editor Engagement Experiments
Start:	2012-12
End:	2013-05-06
Team members:	Ryan Faulkner, Dario Taraborelli

UserMetrics is the name of a platform developed by the Wikimedia Editor Engagement Experimentation (E3) team to measure user activity based on a set of standardized metrics. Using this platform, a set of key metrics can be selected and applied to a cohort of users to measure their overall productivity. The platform is designed for extensibility (creating new metrics, modifying metric parameters) and to support various types of cohort analysis and program evaluation in a user-friendly way. It accepts requests via a RESTful API and returns responses in JSON format.

Changelog

0.1.4-dev

To be released on March 25th 2013.

User Session Management
- flask-login
Implementation of request caching outside the runtime
- start with pickle objects
Client for programmatically accessing API
- https://github.com/rfaulkner/umapi_client
More robust testing.
- added some testing
Interface to load cohorts
Request notifications listener
Update request_manager module to handle larger responses on the queue
Response Handler that builds responses for caching as they finish

0.1.3-dev

Released on March 11th 2013.

Expanded multi-project support
pure JSON responses
expose all parameters in JSON responses
Deal with undefined metric values in a standard way
Make all metrics relative to a user event (e.g. registration) as metrics like survival and threshold. This can be bettet
Refactor API source - break up functionality for easier maintenance and extensibility.
Manage requests via a separate job manager.user_metrics.api.engine.request_manager module.
Generate unique hashes from Request objects. build_key_signature method in user_metrics.api.engine.data

0.1.2-dev

Released on February 26th 2013.

Cross project support, cohorts and requests outside of enwiki are supported
Additional aggregators based on numpy methods (mean, median, max, min)
New script "run_ssh_tunnels" that sets up multiple connections to database hosts on different local ports
Redefined configuration settings to more easily allow for defining new connections and throttling max threads

0.1.1-dev

Released on January 30th 2013, codename "UMAPI".

Initial release
Raw, aggregate, and time-series processing of English Wikipedia user cohorts over various metrics
Metrics: http://meta.wikimedia.org/wiki/Research:Metrics#User_metrics
API implementation in flask that exposes these requests via HTTP urls
API also exposes single user requests

Future Work

Redefine API request entry points:
- run/cohort/..
- run/set/..
- run/user/..
Build filters that allow new cohorts to be generated from requests
OAuth/OpenID integration. This is essential for programmatic access.
HTTPS

Rationale

The metrics API is a project whose aim is to create a way to extract the above metrics in a predictable, consistent, and reproducible way. The code base can be found here and is also pushed to a Gerrit project. This project will soon be replicated to the Wikimedia GitHub account and is currently deployed on the Wikimedia stats cluster. Documentation for the source is also hosted on the stats cluster here. The domain of the API is Wikipedia projects (the current focus is on English WP and is currently being expanded).

Start page of UMAPI.

Documentation

End-user documentation
Metric definition and background research