Skip to content

Conversation

@whitehackr
Copy link
Owner

Summary

  • Fix assignment generation to use correct Mar 1-15 experiment participants instead of random users
  • Replace archived imports with modern assignment methods
  • Add missing assignment metadata to mart model
  • Regenerate overlay orders for proper treatment alignment
  • Switch to JSON upload method with timestamp serialization

Results

  • Proper A/B test structure: 514 control vs 577 treatment users (89% balance)
  • Effect size: 16.8% (target: 20%)
  • Sample size: 1,091 participants (excellent statistical power)

Ready for statistical analysis phase.

- Replace archived assignment import with modern method
- Use live assignment approach for Mar 1-15 experiment participants
- Switch to JSON upload method to avoid pyarrow dependency
- Fix syntax error in experiment_effects.py
- Results: 1,091 properly assigned users vs 1,043 random users
- Add timestamp to string conversion for JSON serialization
- Ensure proper result tracking and summary reports
- Align with experiment_effects.py timestamp handling approach
- Results: Complete overlay generation without JSON errors
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants