Download PowerPoint 프레젠테이션

Survey
yes no Was this document useful for you?
   Thank you for your participation!

* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project

Document related concepts

Mixture model wikipedia , lookup

Transcript
Integrated model management
in the data warehouse era
Daniel R. Dolk
Naval Postgraduate School, CA, USA
EJOR 122 (2000)
2000. 4. 6.
양 영 철
[email protected]
Introduction
Introduction
Integrated Modeling Environment (IME)
Modeling
Data
Warehouse
2
Retrospective of IME
The concept of an IME
3
Modeling Integration
Modeling Integration
Model
Definition
Model Integration
Model
Operation
4
Modeling Integration
4 dimensions
•
•
•
•
Organizational dimension
Definitional dimension
Procedural dimension
Implementation dimension
5
Modeling Integration
Organizational
: Strategic modeling
MANAGEMENT
• An econometric
ACTIVITY
MODEL
marketing model
REQUIREMENTS
• A discrete event simulation manufacturing
Strategic
Model
Mktg->
model
Planning
Integration
Mfg->Dist->
• A transportation Econ->Fin
model
Control
Model
• A pricing model Aggregate
Aggregation
Mktg, Mfg, etc
• A financial model
Operations
Mktg
Mft
Dist
Econ
Fin
Individual
Models
6
Modeling Integration
Definitional
: Schema integration
7
Modeling Integration
Procedural
: Process integration
8
Modeling Integration
Implementation
: OO integrated modeling environment
• CSML (Communicating SML)
– Features
• The basic structured programming constructs of
sequence, selection, and iteration
• Demons
• Embedded SML statements for model definition
• Parallel execution of processes
• Transformation operators to solver data structures
• Embedded SQL statements for data manipulation
9
Retrospective of IME
왜 MMS연구가 활발하지 못한가?
• Lack of external demand
• Paradigm-centric nature of MS/OR
community
• Software development effort
• Complexity of structured modeling
• Theoretical difficulties
• Emergence of the Internet
10
Data warehouses
Overview
특징
?
Data
Warehouse
1. Time-series-based
subsets
of
기업의 의사결정 지원을 위한 주제
중심
적이고
통합적이며 시간성을
비휘
the universe
of an가지는
organization’s
발성 자료의 집합
Operational DB
2. Very Large DB, Contain populations
of data rather than samples
3. Multi-dimensional in nature
4. Conventional data modeling tech.
are not relevant to DW
11
Data warehouses
Environment
•
•
•
•
•
•
Data transformation
Metadata management
Database engine
OLAP and data mining tools
Information delivery system
Data warehouse administration
12
Data warehouses
A modeler’s view
• The largest bottleneck in model building is
more likely to be the data than the model
itself
• The OLAP part of data warehouse is
model-poor
• Data mining is a fertile area for the
application of MS/OR tech. And is currently
being investigated vigorously
13
Decision metric
Performance measurement
• Performance
measurement
– The process of quantifying
the efficiency and
effectiveness of a action
14
Decision metric
Anatomy (1/2)
- Metric %MfcCapacityRealized
MCR(
s.USA
Plant
s
Product
p
p.TV.Color.27in
Time
t
)
t.Year.Quarter.Month
15
Decision metric
Anatomy (2/2)
• The primary attributes of metrics
–
–
–
–
–
–
–
–
–
Definition
Computational procedure
Dimension/units
Thresholds : User Interface 구축에 필요
Periodicity
Scale level
Drilldown dimensions
Data and/or model sources
Report distribution profile
16
Decision metric
Decision metrics in a DSS context
Metric
17
Decision metrics & MM
Decision metrics
as outcomes of models
• Decision metrics may be directly
calculated from the outcomes of math
models
• The associated drilldown allows the users
to specify “what if” analyses by arbitrarily
changing the assigned weights of the
factors and viewing the resulting
evaluations
18
Decision metrics & MM
Models
as predictors of decision metrics
• Models as a means of forecasting future
values of metrics
– A simple trending or exponential smoothing
model
– A more sophisticated approach
• On-line discrete event simulation model
• The shift in modeling emphasis
– Metrics-oriented emphasis on customer
satisfaction vs. traditional operational
efficiency-oriented perspective
19
Decision metrics & MM
Model warehouse
• Stores the information about models
– Model representations (모델 선언부)
– Assumptions
– Interfaces with solvers
• Modeling language statements
Minimize ( TotalCost )
SubjectTo ( testDemand, testStorage,
testInventoryInitial, testInventoryBalance)
For Product = “TV”
Using CPLEX
20
Decision metrics & MM
21
Decision metrics & MM
In order to
Send across
the WEB
22
Component-based IME
Component-based IME
• Intelligent agents
• Component-based software development
– CORBA, COM
23
Conclusions
Conclusions
• Component-based, network-based,
warehouse-based IME
• Distributed System과의 연동
– CORBA, ODBC
• 의문점
– Complex data 처리 문제
• OODB의 이용을 고려…?
24
Reference
References
• Model integration and a theory of
models;D.R.Dolk,J.E.Kottemann;DSS 9 (1993)
• Meta-modeling Concepts and Tools for Model
Management : A Systems Approach;
W.A.Muhanna,R.A.Pick;Management Science 40
(1994)
• Adapting on-line analytical processing for
decision modeling : the interaction of information
and decision technologies;Nikitas-Sprios
Koutsoukis,Gautam Mitra, Cormac Lucas;DSS 26
(1999)
25