Collective Procrastination¶

General Description¶

• Collective Procrastination on GitHub

Operationalization¶

Scale¶

• detection level
  • iteration
• granularity
  • day in iteration
• from
  • iteration start date - if null, end date of previous + 1 day
• to
  • iteration end date - max(due date, last issue closure date)
• ordering
  • chronological

Metrics¶

• iteration duration
  • number of days in iteration
  • (end date - start date) + 1
• issue count
  • number of issues belonging to the iteration
• daily average
  • number of issues to be closed per day on average
  • issue count / iteration duration
• date closed count
  • number of issues actually closed on a particular date
• look-behind count
  • number of issues closed in n1 days before (and including) given date
  • sum(date closed count) since given date - n1 days (excluded) until given date (included)
• look-ahead count
  • number of issues closed in n2 days after (and excluding) a given date
  • sum(date closed count) since given date (excluded) until given date + n2 days (included)

Detection Thresholds¶

• silence
  • look-behind count for any day in iteration < x1 * issue count
• cliff
  • look-ahead count for and day in iteration < x2 * n2 * daily

Experimental Values¶

• n1
  • number of days for look-behind
  • 6
  • justification: big enough to not detect weekends + holidays, small enough to detect anti-pattern on one-week-long iterations
• n2
  • number of days for look-ahead
  • 3
• x1
  • silence steepness
  • 0,1 (10% of overall iteration issues closed in 6 days still counts as silence)
• x2
  • cliff steepnes
  • 2 (cliff needs to be at least twice as steep as the ideal)

Visual Validation¶

• issues burndown
  • rock-edge look - 1 occurance
  • staircase look - multiple occurances
• example
  
  • detected on April 10th
• contra-example
  
  • insufficient silence before cliff

SQL query¶

• attachment "Collective_Procrastination_(for_iteration)_-_SQL.sql"