Lesson 3 — Two Process Views and Strategy
Chapter 14 contains two compatible process views. Do not force them into an identical list.
Nine management activities
- Define Big Data Strategy & Business Needs
- Establish Big Data Environments
- Choose Data Sources
- Acquire & Ingest Data Sources
- Develop Hypotheses & Methods
- Integrate / Align Data for Analysis
- Explore Data Using Models
- Communicate Output to Stakeholders
- Deploy and Monitor
Seven-step Data Science loop
- Define strategy/business need
- Choose sources
- Acquire/ingest
- Develop hypotheses/methods
- Integrate/align
- Explore/model
- Deploy/monitor
The broader management view makes environment establishment and stakeholder communication explicit. The seven-step figure focuses the iterative model-development loop.
Strategy before platform
The first formal activity is not “buy a lake” or “pick an algorithm.” It is to define the problem, expected value, data needs, required timing, governance, and roadmap.
Useful strategy questions: - What problem are we solving? - What decision changes if the analysis works? - Which sources are credible and relevant? - What latency is actually actionable? - What is the impact on existing data structures and core subjects? - What benefit justifies the cost of acquisition, processing, and operations?
Changed fact drill
A vendor proposes real-time architecture because it is technically possible. If the business can act only once daily, the need for extreme low latency disappears. The correct response is to reassess cost/benefit and architecture fit, not automatically select the fastest design.
Source: pp. 473–476, 484–485, 495–497.