Big Data Management (Presentation)
The Three Temporal Dimensions of Data Analytics
The Past
Retrospective View
– What happened?
– Why did it happen?
– Uses historical data
– Delivers static dashboards
The Present
Real-time View
– What is happening now?
– Uses real-time data
– Actionable dashboards
– Alerts
– Reminders
The Future
Prospective View
– What will happen next?
– How can I intervene?
– Uses historical and real time data
– Predictive dashboards
– Knowledge-based dashboards
Business Intelligence vs. Data Analytics
Business Intelligence
Information from processing raw data
Structured data
Simple descriptive statistics
Tabular, cleansed & complete data
Normalized data
Data snapshots, static queries
Dashboards snapshots & reports
Advanced Data Analytics
Discovery, insight, patterns, learning from data
Unstructured & structured data
NLP, classifiers, machine learning, pattern recognition, predictive modeling, optimization, model-based
Dirty data, missing & noisy data, non-normalized data
Non-normalized data, many types of data elements
Streaming data, continuous updates of data & models, feedback & auto-learning
Visualization, knowledge discovery
DAMA International DMBOK2
1. Data architecture management
2. Data development
3. Database operations management
4. Data security management
5. Reference and master data management
6. Data warehousing and business intelligence management
7. Document and content management
8. Metadata management
9. Data quality management
The DMM Model
Level 1
Initial
– Ad hoc, inconsistent, unstable, disorganized, not repeatable
– Any success achieved through individual effort
Level 2
Managed
– Planned and managed
– Sufficient resources assigned, training provided, responsibilities allocated
– Limited performance evaluation and checking of adherence to standards
Level 3
Defined
– Standardized set of process descriptions and procedures used for creating individual processes
– Activities are defined and documented in detail: roles, responsibilities, measures, process inputs, outputs, entry and exit criteria
– Proactive process measurement and management
– Process interrelationships defined
Level 4
Quantitatively Managed
– Quantitative objectives are defined for quality and process performance
– Performance and quality practices are defined and measured throughout the life of the process
– Process-specific measures are defined
– Performance is controlled and predictable
Level５
Optimized
– Emphasis on continuous improvement is based on understanding of organization business objectives and performance needs
– Performance objectives are continually updated to align and reflect changing business objectives and organizational performance
– Focus is on overall organizational performance
– Defined feedback loop between measurement and process change
Source: Peter Ghavami, Big Data Management: Data Governance Principles for Big Data Analytics