Best Practices in Data Partnerships Between Mayor's Office and Academia
- 1. BEST PRACTICES: Data Partnerships Between Government and Academia Homecooked for the SoCal Data Science Conference 2017 By Juan S Vasquez Data Programs Manager LA Office of Finance
- 2. VALUE FOR YOU Better identify & scope partnerships. Invitation from me to you. @juansvas
- 3. TODAY Context Best Practices Round 1 2 Case Studies Best Practices Round 2 Open Q&A @juansvas
- 4. JUAN 101 Data Programs Manager, Office of Finance Data Strategist, Mayor’s Office Supported software LatAm localization effort Digital director for political campaign Advertiser Love running, tattoos & my nonprofit
- 5. JUAN 101 Data Programs Manager: - Tableau - Socrata (*open data*) - Esri ArcGIS online - Oracle tools - 3rd Party Systems - Data Science Fellows - Excel - Stakeholder Engagement - User Interviews - UX champions - Public Speaking
- 6. WHY COLLABORATE? Governments need: - Modern skill sets - Iteration - Human capital Programs & Students need: - Professional experience - Access - Networks @juansvas
- 7. BEST PRACTICES ROUND 1 - Specific Problem Statement - Right Partner - Clear Timeline & Deliverable - Regular Check-ins - Path for continuation @juansvas
- 8. SAMPLE TIMELINE Scope Kick-Off Data Exploration Scheduled Check-ins Delivery Implementation Growth @juansvas
- 10. STORY #1 Workers’ Comp Claim Highest Cost Profiles Cornell University Applied Stats Graduate Program
- 11. STORY #1 Which workers’ comp claims tend to cost the city the most money?
- 12. STORY #1: Highest Cost Profiles
- 13. STORY #1: Highest Cost Profiles Decision Tree @juansvas
- 14. STORY #1: Highest Cost Profiles
- 15. STORY #2 General Assembly Data Science Immersive Finance Website Trends: Public vs City Employees
- 16. STORY #2 General Assembly Data Science Immersive Will inform: - Web content strategy - Call center training materials - Social media strategy @juansvas
- 17. Additional Best Practices Storytelling + actionable insights 1 - 8 month timelines Sweet spot: Scope 1 semester, analyze the next Small, lean team Assign a project manager Details of deliverable matter - format, length, code Keep it simple and focused Address a real, validated problem Pivot if need-be, don’t rush @juansvas