User:Terrrydactyl/Wikimetrics

Here you will find a draft of what I am working on with Wikimetrics. I'm working with Grantmaking to implement features that will help them evaluate programs. Please leave your comments here for now.

Features To-Do

Tagging

We would like to implement a feature that would allow a user to tag a cohort. Tagging cohorts will help users to create cohorts from similar cohorts. For example, if a user would like to look at the retention of active editors from all their March edit-a-thons, they can find those cohorts tagged as March 2014 edit-a-thon to examine all those cohorts and usernames tagged that are appropriate for follow-up at different follow-up times. Further, we would like to allow sharing to the evaluation team so that they can directly find program cohorts to analyze various metrics from particular program implementations for high-level analysis and follow-up inquiry. This will help give an idea of how a program performs in terms of target metrics overall. It’s a way to collaborate with other users to get a bigger picture. Tagged cohorts from other users will only be seen if that cohort is public or shared with the user. There will be specific WMF superusers who will be able to look at all cohorts of a certain tag unless the user chooses to opt out through auto-sharing to WMF admins based on certain types of tags (i.e., programs tags)

For project tracking see Analytics/Wikimetrics/Adding_New_Features/Tag_Cohorts

Delete Wiki Users

Legal requires that any user from in-person events (or possibly any user) be given the opportunity to opt-out of Wikimetrics tracking. Currently there is no simple way of doing that without deleting the entire cohort and deleting the user off application and reuploading the cohort. The delete username is a feature that provides for cohort management to allow for opting-out without the need to revise the complete username list. would need if a wiki user decided to opt out of a cohort. We have to comply with user wishes and take them off the cohort for privacy reasons, however, deleting a wiki user also presents a challenge as reports made from the previous cohort would not be reflect such changes and metrics output may deviate to various extents depending on any omitted users activity.

CentralAuth Cohorts

Sometimes, it's helpful to know about the wiki user's contributions across all their active wiki accounts instead of just one. By querying CentralAuth for user accounts across the projects we will retrieve a cohort list that is exhaustive of all user accounts tied to that wiki username and generate an expanded cohort out of that. There are two ways to do this, on a the cohort creation level and the report running level. It will be best to do it during cohort creation because fetching all the wiki user's user accounts at the time of running a report will take a long time and would have to be redone each time a report is run. However, if it's created at the cohort creation level, the information may be out of date after a certain amount of time. We should also add a way to "refresh" the cohort and recheck all the accounts.

For project tracking see Analytics/Wikimetrics/Adding New Features/CentralAuth Cohorts

Add projects to the CSV file

Since cohorts can be cross wikis, it's helpful to know which wiki the data is being pulled from. Currently, the project is displayed in the JSON output of reports, but not in the CSV output. Many users depend on the CSV output so it's important to have the project listed in addition to the usernames.

Shared Cohorts

[Not an active project] Shared cohorts is a feature that was semi-shelved by Analytics since it wasn't directly related to EEVS. Grantmaking would like the ability to share cohorts as it's an important feature to have in conjunction with tagging. This is an important feature for Grantmaking, but unsure if I'll have the contract time to tackle it.

Completed Features

Delete Cohorts

Wikimetrics did not have a way to delete a cohort completely. Previously, you could delete the cohort from your personal list of cohorts, but not from the database. This meant that the name of the cohort would not be able to be reused and there would be an unlinked cohort floating in the database. Delete cohorts is meant to help delete cohorts that users have made an error on or no longer wish to have.

Add place to paste usernames

Wikimetrics now supports the ability to paste usernames in a text box instead of uploading a CSV file. This helps with some formatting errors that can occur when passing around CSV files and makes uploading cohorts more accessible to people who have an aversion to CSV files.

Add username field in reports

Added a field in the CSV report that shows the username associated with the wiki user. This will be easier to visualize than having just user ids as the identifier. Especially helpful in smaller cohorts where a user may recognize all the wiki users in the cohort.