pharmaR / riskassessment

Collaborative Deployment: https://app.pharmar.org/riskassessment/ Risk Assessment Demo App: https://rinpharma.shinyapps.io/riskassessment
https://pharmar.github.io/riskassessment/
Other
101 stars 28 forks source link

List Decision Automation Rule # in `Decision By` field #682

Open AARON-CLARK opened 11 months ago

AARON-CLARK commented 11 months ago

Right now, the Database Overview doesn't do a good job of maintaining the relationship between the automation rules and the Decision By column seen below. Naturally, we'd want to see the rule responsible for assigning those decisions named in that column, like the current workflow that exists on the Upload Packages tab. However, the relationship is disjointed because after an upload occurs, the decisions can become "stale" if an admin or lead changes the decision automation rules without re-assessing all the pkgs in the database. Therefore, if Rule 1 determined that package xyz was should be "High Risk" at timepoint A, and then an admin changed Rule 1 completely, or perhaps just deleted it, we can no longer point to Rule 1 as being responsible for said decision.

image

As such, I think users should be prompted whenever changing rules if they'd like to "apply decision automation on current database". If not, we can continue to use the tag "auto assigned" here, which is intentionally vague. But if they confirm, we could launch a process that reassigns decisions. What's great is that this process wouldn't need to re-run pkg_ref() or pkg_assess(), or even re-score pkgs, thus preserving those metrics at the original timepoint they were assessed, and avoiding unnecessary compute resources. Instead, the process would unserialize the blob for each pkg and run it against the new rules. If the rules only include score-based rules, we wouldn't even need to unserialize the blob, since those are already stored.

Thoughts @Jeff-Thompson12?

aclark02-arcus commented 3 months ago

I think this is still interesting to look into, but it is not necessary, and thus, a really low priority.