How does Uplevel define Complex PRs?

PR complexity considers lines of code, number of files, and whether it was a refactor. High Complexity PRs may be difficult and take more time to review, while breaking down PR’s into lower complexity can be easier to review and lead to increased velocity and higher release frequency.

PR complexity is based on the probability of being complex to review (on a 0-1 scale).

Uplevel evaluates stuck PRs base on low-medium-high complexity. The stuck calculation and complex or not are for different purposes so in this case they aren’t directly comparableTo calculate being stuck the following probabilities are used:

• low < 0.25
• medium between 0.25 and 0.6
• high > 0.6

The complex / not complex to review is split in half:

• not complex < 0.5 (for stuck this is low + most of medium)
• complex > 0.5 (for stuck this is a bit of medium + high)

Request: { "lines_added": 0, "lines_deleted": 1000, "number_files": 1, "pr_id": "boreas", "pr_description": "rjierpe", "pr_title": "stuf", "jira_description": "dogs", "jira_title": "corndog"

Response: { "probability_of_complex_pr": 0.05930486404317573, "classification_label": "NOT_COMPLEX", }