Big Analytics: Building Lasting Value
0 likes1,251 views
Think Big Analytics, formed in 2010, specializes in launching and scaling big data solutions, providing services such as strategy, engineering, and data science. The company has blue-chip clients across various sectors like banking and retail, focusing on integrated value by understanding business needs and leveraging new data opportunities. Their approach emphasizes deep learning, collaboration across functions, and successful incorporation of data science into business processes.
Big Analytics: Building Lasting Value
- 1. November 2013 BIG ANALYTICS THE GOOD & THE VALUE
- 2. About Think Big Analytics ¨ Formed in 2010 to help clients launch and scale-out Big Data solutions ¨ Services include Big Data strategy, training, engineering and data science ¨ ¨ Management Background: Quantcast, Cambridge Technology, Oracle, Sun Microsystems, Accenture Blue chip clients, including: Ø Internet Transactions Security Global #1 Ø Retail 2 of Global Top 5 Ø Banking 4 of Global Top 1; Financial Services 2 of Global Top 5 Ø Asset Management Global #1 Ø Disk Manufacturing Global #1 Ø Social Networking Global #1 CONFIDENTIAL | 2 2
- 3. Think Big Integrated Value Integrated Value Advisory ¨ ¨ ¨ ¨ Understand true business needs Evaluate suitability of new technologies ¨ Provide perspective on market ideas ¨ Ensure engineering and analytics support business goals Help establish realistic and attainable objectives Drive client-specific innovation Implement ¨ ¨ ¨ Understand technology preferences and limitations Assess talent skills and development needs Develop deep knowledge of the data and tools CONFIDENTIAL | 3
- 4. Big Analytics Ÿ Ÿ Ÿ Ÿ New data Yielding new opportunities Enabled by new approaches With supporting organization CONFIDENTIAL | 4
- 6. Nontraditional Formats Ÿ Unstructured data != text - Call logs - Raw video - Satellite photos CONFIDENTIAL | 6
- 7. Exhaust Ÿ Byproduct data Ÿ Driving interest in the Internet of Things Ÿ Our machines tell a story about us CONFIDENTIAL | 7
- 8. Data about Data Ÿ Data usage patterns Ÿ Driving next generation organizations - Data access patterns as KPI - Systems access as employee engagement CONFIDENTIAL | 8
- 10. Fingerprinting Ÿ Unintentional patterns define us - ATM rhythm - Botnet synchronization Ÿ More connected world exposes more fingerprints - Mobile installs and settings + NFC - Sensory data at shopping mall displays CONFIDENTIAL | 10
- 11. Dark Data Insights Ÿ Ÿ Ÿ Ÿ Ÿ It’s back from the dead! Audit data Fund manager predictions Employee logs Architectural records CONFIDENTIAL | 11
- 13. Unstructured Analysis Ÿ Non-traditional structures - Path models - High dimensionality Ÿ Text - POS - Classification Ÿ Images - Object recognition - Time differentials CONFIDENTIAL | 13
- 14. Deep Learning Ÿ MapReduce built for - Bootstrapped models - Partitioning data by complex logic Ÿ Backpropagation is hard Ÿ Feature learning isn’t (always) CONFIDENTIAL | 14
- 16. Organizational Integration Ÿ Traditionally under engineering Ÿ Integrated with data creators, not data consumers Ÿ Disconnected from business priorities CONFIDENTIAL | 16
- 17. Success Loops Ÿ We take BI for granted - Analysts find novel patterns - Business sees new trends - Statistics is balanced by domain knowledge - Integration of actors aware of feasibility, cost, and impact Ÿ Where does your data scientist sit? CONFIDENTIAL | 17
- 18. Successful Incorporation of Data Science CONFIDENTIAL | 18
- 19. Partnership Ÿ Business is a partner, not a customer Ÿ New insights, capabilities, and products are not born in a vacuum CONFIDENTIAL | 19
- 20. Cross Functional Teams Ÿ Data science is a process, not a job role - Engineering - Research - Statistics - Business - Salesmanship Ÿ Successful Big Analytics blends skills, perspectives, and pushes boundaries CONFIDENTIAL | 20
- 21. Measurement Ÿ Requires KPI/KRI Ÿ Performance metrics - Direct actions - Create purpose CONFIDENTIAL | 21
- 23. Example Client Phase 1 Ÿ First phase: Big Analytics execution Ÿ New methods of Botnet detection Ÿ Led to patent CONFIDENTIAL | 23
- 24. Example Client Phase 2 Ÿ Further analysis - Improvement of Botnet models Ÿ Expansion of cross functional Big Analytic team - Tool selection - Training - Early win identification - Self-selected group CONFIDENTIAL | 24
- 25. Example Client Phase 3 Ÿ Cross-Functional Analytic Organization Ÿ Ÿ Ÿ Ÿ Governance Ownership and accountability Process Roadmap CONFIDENTIAL | 25