Data Science Salon: Building a Data Science Culture
1 like590 views
The document discusses the importance of building a data science culture within organizations, emphasizing that technology alone is insufficient for success. It highlights the need for leadership, team structure, and effective processes to support data-driven work, which ultimately leads to improved decision making and productivity. The document also outlines the importance of training and collaboration across functional teams to address varying business needs.
Data Science Salon: Building a Data Science Culture
- 1. arangocatalina Building a Data Science Culture Catalina Arango Co-founder - Exteractions, Spacewolff, Data.Miami
- 2. arangocatalina “It’s not enough to introduce big data technology, fancy algorithms, and some unicorn-like isolated data scientists to any organization and work culture, and expect them to be able to change things on their own. In order to get the most out of data science, the culture has to change to support data-driven work,” writes Johan Himberg, Data Scientist at Reaktor
- 4. arangocatalina It takes time and hard work to build a data science culture… it’s worth it.
- 5. arangocatalina Why it’s worth it to build a data science culture. Insights gained by data science won’t be implemented without culture Productivity Gains Stat on productivity from analytics Timing Matters for ROI Return on investment is greater for early adopters Widening Gap The gap between leaders and laggards in the data science and AI space will continue to widen
- 6. arangocatalina Data Science Culture People Product DataToolsTeamLeadership Process Business NeedsOrg Structure
- 7. arangocatalina People - Leadership ❏ Have an evangelist or sponsor ❏ Educate management ❏ Always be thinking about the data ❏ Provide intro hands-on training in data science
- 8. arangocatalina People - Team WHO ● Interdisciplinary ● Analytical ● Curious ● Technical ○ R & Python Users ○ Understanding of information systems HOW ● Lateral moves ● Internal Promotions ○ Training ● External hires ○ Grad Students ○ Data Scientists
- 9. arangocatalina Data Science Culture People Product DataToolsTeamLeadership Process Business NeedsOrg Structure
- 10. arangocatalina Process The structure of the data science process will rarely change... UNDERSTAND DATA MODEL DEPLOYCOMMUNICATE … but the organization needs to know how this is actually executed.
- 11. arangocatalina Process - Org Structure Regardless, cross-functional collaboration is critical, especially when domain knowledge validation is needed Centralized function ● More autonomy + ● Isolation from major business functions -- Data scientists found on different teams ● Promotes collaboration with functional teams ● Reduces autonomy Single unit, with specialization ● Works best for large data science teams Stand-alone Embedded Hybrid
- 12. arangocatalina Process - Business Needs Different business needs Different data science deliverables ● Use advanced analytics to improve decision making ● Require strong communication and visualization skills ● Improve products, either internal or external, ex. better search results & recommendations ● More collaboration with engineering Decision Science Data Products
- 13. arangocatalina Data Science Culture People Product DataToolsTeamLeadership Process Business NeedsOrg Structure
- 14. arangocatalina Product - Tools Access Analysis Dissemination Collaboration
- 15. arangocatalina “If you can’t measure it, you can’t fix it!”
- 16. arangocatalina Product - Data Data needs to be... Complete01 Consistent02 Accurate03 Granular04 data science possible and outputs meaningful