Add YTD total to fct_monthly_active_students

[jordan-springer]

Description

Add school_year to date total to rows in fct_monthly_active_students

1. calculate first activity month for each student
2. count distinct students for each first activity month
3. sum the total unique first activities over the window for school year

Links

Jira ticket(s): DATAOPS-1068

Testing story

Example for this school_year, US:

cc @coryamanda

• Does your change include appropriate tests on key columns?
  eg.
  - not_null
  - unique
  - `dbt_utils.unique_combination_of_columns: , ["value","value","value"...]

Note: when submitting a new model for review please make sure the following have been tested:

1. The model compiles (dbt build -m 'your_model')
   or: has the dbt Cloud job succeeded?
2. The model runs (dbt run -m 'your_model')
3. The model produces accessible data in the DW (select 1 from 'your_model')

Privacy

• 1. Does this change involve the collection, use, or sharing of new Personal Data?
• 2. Do these data exist in the appropriate schema(s)?
• 3. Does this change involve a new or changed use or sharing of existing Personal Data?
• 4. Consider: will this data be visible on Tableau? will this data be surfaced in a report exported from Trevor?
• 5. If yes to any of the above, please list the models, columns, and justification below:
  i.
  ii.
  iii.

PR Checklist:

--> Note: if these are not all checked, the PR will be sent back.

• Tests provide adequate coverage
• Privacy and Security impacts have been assessed
• Code adheres to style guide👀 and is DRY
• Code is well-commented (**please do not leave extraneous commentary in model code, if it is for the purpose of documentation, please relocate accordingly)
• Appropriate documentation has been provided (see .yml., did dbt docs generate succeed?)
• New features are translatable or updates will not break up/downstream models
• Relevant documentation has been added or updated (i.e. dbt docs has been updated successfully on Github Pages
• Pull Request is labeled appropriately (eg. chore/, feature/, fix/)
• Follow-up work items (including potential tech debt) are tracked and linked (if applicable)

[coryamanda]

Just for the sake of comprehensiveness - this is double-counting students who use the platform on consecutive months. I need the school year to date numbers for unique students.

So something like:
Month | Country | num_active_students | sytd_active_students
7/1/2024 | US | 1M | 1M <- In July they're equal because it's the beginning of the school year
8/1/2024 | US | 2M | 2.75M <- In August the number of YTD will be lower than the sum of July + August because there will be some duplication
9/1/2024 | US | 4M | 6.25M (edited)
...
6/1/2025 | US | 2M | 9M >- The June school year to date number will be equivalent to the 2024-25 school year-end count of unique students

This is answering this question:
"We're flat October 2024 compared to October 2023. We were 3% down September 2024 compared to September 2023. How are we doing YOY overall?"

[coryamanda]

I found two issues in the previous version:

• It was still double counting students (the sum over partition was on num_active_students, not the num_active_students_ytd field)
• First active month didn't include the school year, so it was taking their first active month ever in the account rather than the first month that school year.

I pushed a commit to fix those two things. It was a breaking change, as my version does not run. There's an error with the last CTE, I think with the partition formatting. But the logic up in the CTEs above (up through final) should now be working, so hopefully the issue in the rolling_final an easy fix. Can you look at this again and see if you can figure out the partition logic in the last step?

[coryamanda]

great - thanks!