Hi all,
We're going to start being more transparent with the community today about the work being done by the analytics development team. Today's update is a summary of our sprint planning that we did yesterday.
Short background -- the development team works in 2 week sprints which usually start on Thursday and end on Tuesday. Wednesday is a day that can be used at the team member's discretion -- normal work can be done, but it's also a space for experimentation.
Over the coming weeks, we'll share more information about our planning, showcasing and tools.
You may be "surprised" at the relatively small amount of work we commit to at the onset of a sprint. This is for two reasons -- we are interrupted a lot and this makes it difficult for us to predict our output; and we are a new team and are still understanding our velocity. Of course if we finish our commitments early, we just grab another task off the top of the backlog and work on that.
Our projects are documented in our Mediawiki prioritizationhttps://www.mediawiki.org/wiki/Analytics/Prioritization_Planning page. The following tasks and features roll are all associated with a prioritized project.
Here are the commitments for this sprint. Links are to Mingle, the agile planning tool that we use.
Fundraising Visualization Syncuphttps://wikimedia.mingle.thoughtworks.com/projects/analytics/cards/1449 Measuring ULSFO Site Performancehttps://wikimedia.mingle.thoughtworks.com/projects/analytics/cards/1450
Here is the top of the backlog (that is the tasks that will be worked on if/when the above are finished)
Productionize Wikimetricshttps://wikimedia.mingle.thoughtworks.com/projects/analytics/cards/1185 Update Wikimetrics Serializationhttps://wikimedia.mingle.thoughtworks.com/projects/analytics/cards/1396 Calculate Mobile Browser Breakdownhttps://wikimedia.mingle.thoughtworks.com/projects/analytics/cards/1227 Duplicate WIkipedia Zero Dashboard Creation on Hadoophttps://wikimedia.mingle.thoughtworks.com/projects/analytics/cards/1453 Run Scheduled Reports via Wikimetricshttps://wikimedia.mingle.thoughtworks.com/projects/analytics/cards/1376
Please let me know if you have any questions.
-Toby
On Feb 21, 2014 10:03 PM, "Toby Negrin" tnegrin@wikimedia.org wrote:
Over the coming weeks, we'll share more information about our planning,
showcasing and tools.
Great, thanks.
Here are the commitments for this sprint. Links are to Mingle, the agile
planning tool that we use.
Fundraising Visualization Syncup Measuring ULSFO Site Performance
Here is the top of the backlog (that is the tasks that will be worked on
if/when the above are finished)
Productionize Wikimetrics Update Wikimetrics Serialization Calculate Mobile Browser Breakdown Duplicate WIkipedia Zero Dashboard Creation on Hadoop Run Scheduled Reports via Wikimetrics
Which of these can be done by people outside the team?
I guess it may make sense to leave the team to do the current sprint and have others work on the backlog where possible? (vs. having everyone just do the current sprint items)
I just clicked on one task that looked interesting: "Duplicate WIkipedia Zero Dashboard Creation on Hadoop". Its unclear how much access is required to do that task. stat1 is enough? is either the legacy sampled logs/dashboard or the new unsampled hadoop infra deployed in labs?
-Jeremy
Productionize Wikimetrics Update Wikimetrics Serialization Calculate Mobile Browser Breakdown Duplicate WIkipedia Zero Dashboard Creation on Hadoop Run Scheduled Reports via Wikimetrics
Which of these can be done by people outside the team?
Technically any of them. A volunteer could sign an NDA and have the same access team members do. But without an NDA, the most obvious tasks available to anyone are anything Wikimetrics. It was done on top of labs db partly for this reason.
I guess it may make sense to leave the team to do the current sprint and have others work on the backlog where possible? (vs. having everyone just do the current sprint items)
If someone's interested in work, just talk to me (milimetric on IRC). I'll help coordinate and find you something you like doing.
I just clicked on one task that looked interesting: "Duplicate WIkipedia Zero Dashboard Creation on Hadoop". Its unclear how much access is required to do that task. stat1 is enough? is either the legacy sampled logs/dashboard or the new unsampled hadoop infra deployed in labs?
For that one, someone would need access to the hadoop cluster, so analytics10**.eqiad.wmnet and knowledge of the current wikipedia zero dashboard process (which is a few python scripts and some shell scripts). But feel free to ask me about any of the tasks on IRC or here.