2. Agenda or TOC
• The Analytical
• Data
, where there is no data to analyse to
where we have too much of data to analyse
• “
” used till date for this
•
we have a single structure for OLTP
and Analytics?
• The ‘data’
of data
• Data
3. Analytical Processing of Data
Operational
Reporting /
MI
Analytics
OLAP / BI / ETL
Descriptive (Uni
or bivariate)
Analytics
Diagnostic or
Inquisitive
Content
(Unstructured)
Discovery
Predictive
Structured
Predictive
Statistical Techniques
Machine Learning
4. Data Scenarios…
• New product design
• Simulation
• Knowledge
representation
No Data
Structured
Data
• From normalized
OLTP systems
• Variables , mostly
numbers
• Unstructured
• Quickly varying
• Mostly non-numeric
BIG data
5. Data analysis technology “names”
• Adhoc Data
& Queries
• Management Information
, Business Intelligence
• Business Analytics
• Real-Time BI
• Artificial
Analytics
• In-memory
6. Why can’t we analyze OLTP data
•
•
•
•
OLTP schema
Optimized for handling
transactions i.e., updates
Short and quick transaction
are catered
Multiple concurrent users
read small amounts of data
Key based, index lookups
used to access data
•
•
•
•
OLAP schema
Optimized for large loads of
data in ETL mode
Large summarizations to be
performed
Less number of users read
huge amounts of data
Full scans through data are
often required
9. Business Value
Business Value - Analytics Matrix
What is the best that can happen?
What will happen?
Optimization
Linear/Non-linear
programming & Simulations
Predictive Modeling
Baseline Demand
Impact of Causal Factors
Descriptive Modeling
Why something happened?
Describe historical event
Insights/Limited What-if
A
n
a
l
y
t
i
c
s
Actionable insights
What happened?
R
T
B
I
OLAP Reporting
Drill-thru
Drill-Across
Standard Reporting
Sales, Inventory, Business
Performance
Data Management
Internal, Syndicated,
Decision Support
DSS
Decision Guidance Advanced analytics
DSS – Decision Support Systems,
RTBI – Real Time Business Intelligence
9