Analyze AY2024 cognizant assignments

[danswick]

We need to understand the size and shape of AY2024 audits that should have been given a new base year cognizance assignment for 2024. Possible next steps include:

1. querying our data for audits that meet the criteria: audit year is 2024, cognizance has been assigned.
2. for the audits identified in the previous step, ignore the prior base year and recalculate cognizance.
3. Compare the cog assignments from both 1 and 2 above and generate a list of audits where they differ.

[gsa-suk]

@danswick Analysis details are here - https://drive.google.com/drive/folders/1qFchSD-3P3zFT0rw6L87YGqz0F6flTQQ

[danswick]

Thanks, @gsa-suk! I'm copying the summary below so it's easier to find later. I tagged you in a few places where we could use a little more detail. Please correct me if I'm wrong, but I think the conclusion we should take from this exercise is something like:

we only used a 2019 base year for 11 of the cog assignments. The other 58 had to be calculated. Each of those 11 base year examples would have had the same assignment if we had used 2024 as a base year.

Is that correct?

---

Analysis of cog assignments for audit_year 2024

Purpose:
In the production application, cognizant assignments are calculated with the baseline year of 2019. For the 2024 submissions i.e., submissions where audit year=2024, we want to know if cog assignments would be different if the baseline year had been based on ‘2024’ instead of ‘2019’.

Data used for analysis:
Prod support_cognizantbaseline, support_cognizantassignment, dissemination_general and dissemination_federalaward tables obtained on 10/29/24.

Code used for analysis:
https://github.com/GSA-TTS/FAC/tree/sk/cog_2024_analysis. Overview of changes in this branch:

1. cog_over.py was modified to accept a base_year parameter. BASE_YEAR was set to ‘2024’ for this analysis.
2. check_cog_for_2024.py management command was created. [@gsa-suk what does this management command do?]

Results of running the above code:

• Total # cognizant assignments in this data: 69
• Out of the above... Total # cog in the support_cognizantbaseline table: 11
  • [@gsa-suk what does this mean?)
• Total # calculated cog assignments \= 58

Findings:

• Using the support_cognizantbaseline table per the cog/over flowchart:
  • 58 cog assignments were calculated by GSAFAC using the 2024 base year. These are identical to those generated with the 2019 base year. This was found out by comparing the new calculated value to the value present in dissemination_general. 📄 cog_Prod_2019_cog_2024_w_baseline.xlsx
  • 11 cog assignments were present in the support_cognizantbaseline table and hence not re-calculated by the management command.
• Skipping the support_cognizantbaseline data and re-calculating cog for all 69 audits:
  • All cog values calculated with 2024 base year are identical to those calculated with 2019 base year. 📄 cog_Prod_2019_cog_2024_skip_baseline.xlsx

[gsa-suk]

@danswick The 11 entities that have cog in support_cognizantbaseline table do not have 2019 audits in the dissemination_general table or had oversight in 2019. GSAFAC calculated cog from current data and stored in the baseline table in various dates after MVP release.

[danswick]

In that case, we might need to revisit the summary above using more descriptive terms instead of the table names (support_cognizantbaseline, etc).

The final bullet point kind of makes all of this moot, but I want to make sure your findings are reflected accurately for future reference.

[gsa-suk]

Summary: Using 2019 base year or 2024 base year for cog calculations of 2024 audits in Prod data from 10/29/24 yields the same result.