Create the "fact tables" for our metrics in data/metric-definitions #5970
Description
Developer Metrics Definitions
Activity
data/metric-definitions/activity.py
Scope
Define canonical measures of developer activity and thresholds for determining “active” status.
Metrics
- Commits: count of distinct days with at least one commit
- Push events: count of distinct days with at least one push event
- Issues / PRs: count of distinct days with issue or pull request activity
All activity metrics are expressed in terms of distinct active days, not raw event counts.
Activity Thresholds
- Electric Capital definition
- Active if ≥10 distinct commit days within a 28-day rolling window
- OSO definition
- Active if ≥10 distinct days with any of: Commit, Push Event, or Pull Request activity within a 30-day rolling window
Outputs
- Boolean activity flags per developer per month:
- is_active_electric_capital
- is_active_oso
- Supporting metrics:
- active_days_commits
- active_days_pushes
- active_days_issues_prs
- active_days_any
Goal
It should be possible to produce line charts comparing Electric Capital MAU and OSO MAU over time, with trends that are directionally consistent and ideally closely overlapping.
Lifecycle
data/metric-definitions/lifecycle.py
Scope
Define developer lifecycle stages and capture all valid state transitions over time.
Lifecycle Stages
- New developer
- Full-time active developer
- Part-time developer
- Dormant
- Churned
Each developer must be assigned exactly one lifecycle stage per time period.
Transitions to Capture
- New developer → Full-time active developer
- Full-time active developer → Part-time developer
- Part-time developer → Full-time active developer
- Any active state → Dormant
- Dormant → Active (full-time or part-time)
- Any state → Churned
Outputs
- Lifecycle stage per developer per month
- Explicit transition records:
- developer_id
- from_state
- to_state
- transition_month
Goal
It should be possible to generate bar charts showing the distribution of developers across lifecycle stages using either ODD- or OSO-derived activity metrics, with comparable aggregate patterns.
Alignment
data/metric-definitions/alignment.py
Scope
Measure how a developer’s activity is distributed across ecosystems within a given time interval.
Definition
Given:
- A predefined list of ecosystems
- A developer
- A time interval (e.g., calendar month)
Compute the percentage of the developer’s total activity attributable to each ecosystem during that interval.
Percentages must sum to 100% per developer per time period.
Example
- March 2025:
- Ethereum: 70%
- AI: 30%
Outputs
- developer_id
- ecosystem
- time_period
- activity_share (0–1 or 0–100)
Retention
data/metric-definitions/retention.py (TBD)
Scope
Measure developer retention over time within a given ecosystem.
Definition
For each (developer, ecosystem) pair:
- Define Month 0 as the month of first observed contribution
- Track activity presence for Month 0 through Month N using OSO activity definitions
Outputs
- Cohort tables keyed by first-activity month
- Retention counts or rates at each month offset
- Optional boolean flags for active vs inactive per offset month
Goal
It should be possible to generate cohort tables and line charts showing developer retention curves over time for each ecosystem.