...
- How to capture the
attempts? i.e. Number of times the particular user is attempted particular question. - How to capture the
batch-idandcourse-id
Conclusion:
Analytics team:
1. Analytics team will store the attempts as a blob in the database and all the event data related to the questions will be stored in the blob.
2. Analytics team will implement a API to ingest assessment related data. The API will take course_id and batch_id and a batch of ASSESS events.
3. The API will route the events to a separate Kafka topic. A new Samza job will process these events and load the summarized data into the database. Each record will correspond to a attempt_id for a worksheet. The record will also contain overall score for the attempt_id.
4. Postgres might not scale for the number of ASSESS events every day. Cassandra will be used to store the summarized data.
Portal/Mobile (Estimations have not been accounted for the following tasks):
1. Currently, the question id is auto generated every time the worksheet is played. Content player needs to fix it and use the do_id for the questions.
2. Mobile and portal will have to send the attempt id in cdata both in the case of practice questions and exams. Currently, a new attempt id can be generated every time the worksheet is played. However, in future the exams will need to have the same attempt id passed until the assessment is submitted.
3. Mobile/Portal should figure out a way to call the assement score computation API only for assessments worksheets.