Notes from Ali Swanson on how to get started duplicating the aggregate report

• Brooke has some code at

  • https://github.com/zooniverse/Data-digging/blob/master/example_scripts/wildwatch_kenya/aggregate_survey.py
  • that runs basic aggregation on the raw data.

• The official aggregation approach and code for Snapshot Serengeti (e.g. www.nature.com/articles/sdata201526) is here:

  • https://github.com/mkosmala/SnapshotSerengetiScripts

• The Oxford team is working to fix aggregation through the project builder - I'm hopeful we might be able to have something built this summer, but there are no guarantees.

Report types, basic

• workflow contents - has workflow id, contents version, and question choices
• workflows - has workflow id, workflow version, tasks, and some other stuff I don't understand fully
• subjects - has subject, project, workflow, subject_set IDs, image names, image links, classification count, retirement date/time, reason
  • this is the one to look at to see what has not been retired yet
  • could be used to see what is currently linked to workflow
• classifications - included user information and ip, workflow_id and version, created at date/time, session information, classification answers, subject data and ids

Old aggregation report

*

Todo

• need to find a way to recreate the aggregate report from either the classifications report
• No clue where to even start with this