1 Introduction CIM or/and DM Rong-Ho (Ron) Lin, Ph. D. Visting Researcher in the University of Iowa E-mail: [email protected]Associate Professor National Taipei University of Technology Institute of Commerce Automation and Management, Department of Business Management E-mail: [email protected]CIM or/and DM Data Mining…Management Manufacturing…Services Computer Integrate…Interface Manufacturing…Production Outline CIM Brief Overview What is CIM? … DM Brief Overview What is DM? … CIM and DM Relationships Research Areas in CIM and DM Software Demo. CART 4.0 for Windows XpertRule Miner
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
1
Introduction CIM or/and DM
Rong-Ho (Ron) Lin, Ph. D.
Visting Researcher in the University of IowaE-mail: [email protected]
Associate ProfessorNational Taipei University of Technology
Institute of Commerce Automation and Management,Department of Business Management
Ø CIM and DMð Relationshipsð Research Areas in CIM and DM
Ø Software Demo.ð CART 4.0 for Windowsð XpertRule Miner
2
What is CIM ?
Ø a computer system? Ø a management philosophy?
Ø a competitive strategy? Ø a data management and networking problem?
Ø a concept? Ø a technology?
Ø a method for eliminating direct labor? Ø a deterministic technical system?
Ø a socio-technical system?
Definition of CIM
Computer Integrated Manufacturingis the integration of the total manufacturing enterprise through the use of integrated systemsand data communications coupled with newmanagerial philosophies that improve organizational and personnel efficiency.
*from the Computer and Automation Systems Association (CASA) of the Society of Manufacturing Engineers (SME).
Manufacturing Enterprise (1)
•GOAL : stay in business, grow, and make profit•1970s: COST•1980s: QUALITY and COST•1990s: QUALITY, COST, TIME, plus customization…•Globalization à Dynamic Market à Flexibility•What can we do under these changing circumstances to meet the challenges of present and future markets in order to stay in business, to grow, and to make profit
•Virtual Enterprise à Global Manufacturing: optimum combination of resources of production available throughout the world…
3
Manufacturing Enterprise (2)
MANAGEMENT
DESIGN PRODUCTION
MARKET
• Concurrency among all decision making levels• Timeliness of information
Major Subsystems in CIM
MARKETING
CA-Storage & Transportation
(Logistics)
INFORMATION SYSTEM
QUALITY
Information & Material Flow
(Lower Level Controllers)
CNC, PLC, ROC, AGV,…
CAMSFC CAPC TMS QC
PPS(Aggregate Plan., MRP)
CAPP
ProductCAD
Tool Machine
CIM Wheel -1985
OUTER RING: Business Segments
MIDDLE RING: Design & Mfg
CENTER: Information System
4
Enterprise Wheel - 1993
MIDDLE RING: Business Design
OUTER RING: Business Global
CENTER: Customer Satisfaction
INNER RING: Knowledge Managt.
Historical Background (I)
1970s - 1980s…Flexible Manufacturing Systems/Cells •“… a group of CNC machinery/equipment linked by an automated materials handling system, whose operation is integrated by supervisory computer control …”
•Provided efficient mid-volume, mid-variety production
•Increased machine utilization•Reduced work-in-process inventory, labor and tooling costs, setup costs, lead time, lot size
•Ability to react quickly to engineering changes
Historical Background (II)
1980s...Computer Integrated Manufacturing•“… computer assisted control and integration at all levels of a manufacturing enterprise …”
•Strict hierarchical information flow•Totally automated, unmanned, paperless factory
•Integration problems in heterogeneous environments
5
Historical Background (III)
1990s…Intelligent Manufacturing•“… architecture for highly decentralized manufacturing systems, built from a modular mix of autonomous, cooperative, and intelligent elements (agents) …”
•Re-emergence of the human factor seeking tomaximize the use of human intellectual skills and flexibility
•Migrating from hierarchy to Vertically•Capability of rapid self-reconfiguration by incorporating human and machine intelligence
Historical Background (IV)
1990s…Extended Enterprises• “… global integration throughout the complete supply chain in
manufacturing: Virtual Enterprise …a community of factories each focused on what it does best, all linked by a network that would enable them to operate as one flexibly and inexpensively regardless of their location”
• An enterprise mostly made of functions provided by other enterprises
• Extensive use of standards and information technology tools• Organizations and teams operate autonomously but
cooperatively• Pushing toward true integration of design and manufacturing
• Definition of Data Mining• Definition of Data Mining
• Data Mining is defined as the process of extracting valid, previously unknown, comprehensible information from large databases in order to improve and optimize business decisions.
DM Successes Applications
• Decision trees constructed from bank-loan histories to produce algorithms to decide whether to grant a loan.
• Patterns of traveler behavior mined to manage the sale of discounted seats on planes, rooms in hotels, etc.
• Diapers and beer ." Observation that customers who buy diapers are more likely to by beer than average allowed supermarkets to place beer and diapers nearby, knowing many customers would walk between them. Placing potato chips between increased sales of all three items.
• Skycat and Sloan Sky Survey: clustering sky objects by their radiation levels in different bands allowed astromomers to distinguish between galaxies, nearby stars, and many other kinds of celestial objects.
• Comparison of the genotype of people with/without a condition allowed the discovery of a set of genes that together account for many cases of diabetes. This sort of mining will become much more important as the human genome is constructed.
The Data-Mining Communities
• Statistics.• AI, where it is called “machine
learning."• Researchers in clustering algorithms.• Visualization researchers.• Databases.
8
Features of Data Mining
• Complex models (AI, ANN, Rule-based)• Large problems (Datasets)• Many discrete variables (Mixed of
continuous and discrete in practice)• Wide use of cross-validation (DT:
misclassification rate)• Few comparisons with simple statistical
models (Difficult to explanation)
Tools of Data Mining
• Data warehouses, marts, process, tasks• association, clustering, classification, forecast, • methods • statistics, decision trees and rules, neural networks, fuzzy learning,
and case-based reasoning• machine learning in design• knowledge acquisition, learning in analogical design, conceptual
design, and learning for design reuse
Stages of the Data-Mining Process
• Data gathering, e.g., data warehousing, Web crawling.• Data cleansing: eliminate errors and/or bogus data, e.g.,
patient fever = 125.• Feature extraction: obtaining only the interesting
attributes of the data, e.g., “date acquired" is probably not useful for clustering celestial objects, as in Skycat.
• Pattern extraction and discovery. This is the stage that is often thought of as \data mining," and is where we shall concentrate our effort.
• Visualization of the data.• Evaluation of results; not every discovered fact is useful,
or even true! Judgement is necessary beforefollowing your software's conclusions.
9
CIM and DM
“Enterprise Integration”
Knowledge Management
(Cooperative Information Systems Integration)
SCI
SCM ERP CRM
Manufacturing Information Systems
Decision Rules
CIM à Information System in CIM
CIM/DM Research Areas
• Data mining for product development and concurrent engineering;
• Design and manufacturing warehousing; • Computer-integrated manufacturing (CIM) and data mining; • Data mining for Material Requirements Planning (MRP); • Enterprise Resource Planning (ERP) and Workflow
Management;• Process and quality control; • Process analysis; • Data representation/visualization;• Fault diagnosis; adaptive schedulers; • Learning in robotics.
Data Mining for Product Development
• Enterprise Learning (For example: Integrated systems and technol ogies; • Knowledge sharing; Enhancing service exchange networks) • Data Mining for Concept Development (For example: Extracting pat terns from
customer needs; Learning interrelationships between customer needs and design specifications; Clustering of design concepts; Indexing and retrieval of design concepts in knowledge bases; Data mining procedures for concept selection)
• Data Mining for System-level Design • Prediction of Product Development Span Time and Cost • Project Evaluation • Visualizing Relationships in Large Product Development Databases• Concurrent Engineering (For example: Extracting interrelationshi ps between
design requirements and manufacturing specifications; Integratio n of data mining and team decision support; Exploring tradeoffs between ov erlapping activities and coordination costs; Combining expert knowledge)
• Marketing and Logistics (For example: Supply and delivery forecasting; Learning suppliers, customers, partners involved in transportation and distribution; Time Series Analysis with Neural Networks for Inventory Applications)
10
Data Mining for Engineering Design
• Data Mining Support for Design from Physical Principles• Extracting Guidelines and Rules for Design-for-X
(manufacturability, assembly, economics, environment)• Extracting Product Characteristics from Prototypes• Case-based Reasoning in Conceptual Design • Clustering of Design Cases for Design Reuse • Interactive Exploration for Conceptual Design• Design Knowledge Acquisition with Data Mining • Design Analysis • Materials Data Mining • Dynamic Indexing and Retrieval of Design Information in
Knowledge Bases • Creative Design using Genetic Algorithms and Evolutionary
Programming• Cost Evaluation Systems (design for assembly and manufacturing)• Industrial Design• Integrated Data Mining and Design of Experiments
Data Mining for Manufacturing
• Selection of Materials and Manufacturing Processes• Time Series Analysis and Data Mining • Fault Diagnosis • Data Mining for Preventive Machine Maintenance • Manufacturing Knowledge Acquisition with Data Mining• Process and Quality Control • Predicting Assembling Errors • Process Analysis • Operational Manufacturing Control (For example: schedules that learn, the effect of local
dynamic behavior on global outcomes)• Dynamic indexing and retrieval of manufacturing information in knowledge bases summarization
and abstraction of large and high -dimensional data (Self Organizing Map (SOM) based data visualization methods)
• Adaptive Human-Machine Interface for Machine Operation• Feature Selection and Dimensionality Reduction of Manufacturing Data • Extracting Process Yield Classes with SOM • Feature Recognition with SOM and Neural Networks• Cutting tool- state classification for tool condition monitoring• Data Mining for Capturing Best Manufacturing Practice • Learning in the context of Robotics (For example: navigation and exploration, mapping,
extracting knowledge from numerical and graphical sensor data) • VLSI implementation of Neural Networks and Fuzzy Systems
Integrating Data Mining Systems
• Deploying Data Mining in On-line Systems• Data Mining for Enterprise Resource
Planning and Workflow Management• Adaptation of Manufacturing Environments
for Data Mining Applications• Infrastructure for Data Mining in
Manufacturing (For example: integrating data mining with Computer Integrated Manufacturing, MRP, and CAD/CAM)
11
Researchers in CIM/DM – (1)
• Choonjong (CJ) Kwak,PurdueUniversity• Data Mining Approach to Production Control in the
Computer Integrated Testing Cell• A data mining-based production control approach for the
testing and rework cell in a dynamic and stochastic Computer Integrated Manufacturing (CIM) system.
• A decision tree-based module generates symbolic classification rules on each partitioned data that are suitable for interpretation and verification by human users and saves the rules in the CDS knowledge bases.
Researchers in CIM/DM – (2)
• Richard G. Mathieu (Rick), St. Louis University– “Kanban Setting Through Artificial Intelligence: A Comparative
Study of Artificial Neural Networks and Decision Trees”, Integrated Manufacturing Systems: The International Journal of Manufacturing Technology Management, Vol. 11, No. 4, 2000, pg. 239-246 (with I. Markham and B. Wray).
– "A Rule Induction Approach for Determining the Number of Kanbans in a Just-In-Time Production System", Computers and Industrial Engineering, Vol. 34, No.4, 1998, pp. 717-727 (with I. Markham and B. Wray).
– "A Methodology for Large-Scale R&D Planning Based on Cluster Analysis", IEEE Transactions on Engineering Management, Vol. 40, No. 2, 1993, pp. 283-292 (with J.E. Gibson).
Researchers in CIM/DM – (3)
• Gautam B. Singh, Oakland University – Data Mining and Predictive Modeling in Manufacturing Databases
• His research in this area aims at developing deterministic and stochastic techniques for learning data models from scientific databases. Specifically, time series method is being applied to find patterns and forecasting trends in the data gathered from amanufacturing assembly line. This information can be utilized toestimate the process capability and improve efficiency.
12
The Sequel to the Internet Revolution
0.1
1
10
100
1,000
19701960 1980 1990 2000 2010 2020
Early adopters
E-mailTCP/IP
WWW
HTML
E-commerce
XMLWAP
Everyone
MilitaryUniversities
Specialists
Inte
rnet
use
rs in
mill
ions
Users
What is CART™? A brief overview
• Classification And Regression Trees (CART)
• Origins in research conducted at Berkeley & Stanford 1984 – Leo Breiman, University of California, Berkeley
– Jerry Friedman, Stanford University– Charles J. Stone, University of California, Berkeley
• The split criterion is based on the minimum value of the Gini Index of the split.
• Where P(j|t) is the relative frequency of class j at node t. • k is the number of children nodes. • ni is the number of records at child i and • n is the number of record at nodes P.
∑j
tjP 2)]|([
)(1
iGINInnk
i
i∑=
First Iteration of GINI Index & Split
Outlook Temperature Humidity Windy
Feature Sunny+ Rain
Overcast Hot+ Cool
Mild High Normal False True
Yes 5 4 5 4 3 6 6 3
No 5 0 3 2 4 1 2 3
GINI(i) 0.5 0 0.47 0.44 0.49 0.25 0.38 0.5
GINIsplit 0.36** 0.46 0.37 0.43
14
Solution from GINI
IF Outlook = Overcast Or Windy = False And Humidity = Normal And Outlook = (Sunny || Rain)OrOutlook = Sunny And Windy = True And Humidity = NormalThen Play Tennis
TerminalNode 1N = 4
TerminalNode 2N = 3
TerminalNode 3N = 1
TerminalNode 4N = 1
Node 4OUT
N = 2
Node 3WIN
N = 5
TerminalNode 5N = 5
Node 2HUM
N = 10
Node 1OUT
N = 14
Overcast Rain/Sunny
Normal High
False True
Rain Sunny
Conclusion:No one tool is best for all problems
• Best tool depends on the nature of the data and the patterns we are trying to discover– Only common broad-scale structures– Multiple structures for different subsets of
data• Best tool depends on quality of the data
– How prevalent are missing values?– How often are fields in error?– Are the target variables ever misrecorded?
How frequently?
• Best tool depends on how long we can afford to wait for an answer– expert resources available or not
Age Income $pend Credit
? ?
?
?21900
46200
Bad Value Bad Value
Bad Value
32
21
16
29 74300
895
35
176
610
490
530
References
• Breiman, L., J. Friedman, and R. Olshen , and C. Stone (1994), Classification and Regression Trees, Pacific Grove: Wadsworth.
• Friedman, J. H. (1991a), Multivariate Adaptive Regression Splines (with discussion), Annals of Statistics, 19, 1- 141 (March).
• Michie, D., D. J. Spiegelhalter, and C. C. Taylor, eds(1994), Machine Learning, Neural and Statistical Classification , London: Ellis Horwood Ltd.
• Steinberg, D. and Colla , P. L., (1995), CART: Tree-Structured Nonparametric Data Analysis, San Diego, CA: Salford Systems.