Tuesday, August 29, 2006
Decision Support System – Prof. Chandan Bhattacharya
MANAGEMENT INFORMATION SYSTEM
The study of Management Information System started in 1960 to focus on computer based information system aimed at managers. MIS combines theoretical work of computer science, management science and operations research. It is a practical orientation towards building systems and applications. This is shown in the following figure.
Diag DSS 1
The technical approach of information system emphasizes mathematically based models as well as the physical technologies to study information systems. The discipline that contributes the technical approach are computer science, management science and operation research computer science is concerned with establishing theories of compatibility, methods of computations and methods of efficient data storage and accesses Management Science emphasizes the development of models for decision-making and management practices. Operation Research focuses on mathematical techniques for optimizing selected parameters of organizations such as transportations inventory control.
The growing part of the information system is concerned with behavioral problems and issues. Sociologists focused on the impact of the information system on organizations and society.
Political Science investigates the political impact and uses of information system and technology psychology concern with individual responses and information systems and cognitive model of human reasons.
DIFFERENT KINDS OF INFORMATION SYSTEMS
There are 4 main kinds of information system serving different organisation levels: -
1. Operations Level
2. Knowledge Level
3. Management Level
4. Strategic Level
This is shown in following diagram: -
Diag DSS 2
Organisations and information systems can be divided into strategic management, knowledge and operational levels. They can be divided further into five functional areas: -
1. Sales & Marketing
5. Human Resources
1) STRATEGIC LEVEL SYSTEM help senior managers for long term planning.
2) MANAGEMENT LEVEL SYSTEM helps middle managers monitor and control.
3) KNOWLEDGE LEVEL SYSTEM help knowledge and data workers design product distribute information and cope with papers work.
4) OPERATIONAL LEVEL SYSTEM help operational managers to keep track of organisations day to day activities.
DIFFERENT TYPES OF INFORMATIONS SYSTEM
6 major types of information system needed for the 4 levels of an organisation as shown in the following table. Information systems are built from each of the four levels of an organization.
A Table Chart
TRANSACTION PROCESSING SYSTEM
Computerised systems that perform and record the daily routine transaction to necessary conduct the business. It serves the operational level of an organisation.
KNOWLEDGE WORK SYSTEMS
Information Systems that aid knowledge workers in the creation and investigation of new knowledge of the organisation. It serves the knowledge level of an organisation.
OFFICE AUTOMATION SYSTEM
Computer Systems, such as word processing, electronic mail etc. are designed to increase the productivity of data workers in the office. It serves the knowledge level of an organisation.
DECISION SUPPORT SYSTEM
Information Systems at the management level of an organisation that combine data and sophisticated analytical model to support semi-structured or un-structured decision-making. It serves the management level of an organisation.
MANAGEMENT INFORMATION SYSTEM
Information System at the management level of an organisation that serves the function of planning, controlling and decision making by providing routine summary and exception reports. It serves management level of organisation.
EXECUTIVE SUPPORT SYSTEM
Information Systems at the strategic level of an organisation designed to address unstructured decision making to advanced graphic and communication.
CHARACTERTISICS OF MANAGEMENT INFORMATION SYSTEM
1. MIS support structured and semi-structured decision at the operational and management control useful for planning purpose of senior management stuff.
2. MIS are generally reporting and control oriented. They are designed to report on existing operations and therefore to help to provide day-to-day control of operations.
3. MIS relay on existing corporate date and data flow.
4. MIS have little analytical capabilities.
5. MIS generally aid in decision making using past and present data.
6. MIS are relatively flexible.
7. MIS have an internal rather than external orientation.
8. Information requires are known and stable.
9. MIS requires a lengthy analysis and design process.
CHARACTERISTICS of DECISION SUPPORT SYSTEM
1. DSS offer users flexibility and adoptability and a quick response.
2. DSS allow users to initiate and control the input and output.
3. DSS operate with little or no assistance from professional programmers.
4. DSS provide support for decisions and problems whose solutions cannot be specified in advance.
5. DSS use sophisticated analysis and no modeling.
Diag DSS 3
The figure above illustrates how various types of system in the organisation are related to each other – TPS are the major producer of information that is required by the other systems which in turn produces information for other systems. These different types of systems are loosely coupled in most organisations.
Decision Support System – Prof. Chandan Bhattacharya – 20th June, 2006.
The non-trivial extraction of implicit, previously unknown & potentially useful information from network.
2) Data summarisation
3) Learning classification rules
4) Analyzing changes
5) Detecting anomalies
1) Data Mining is the analysis of data & the use of software techniques for finding patterns & regularities, in sets of data.
2) It is responsible for finding the patterns by identifying & underlying rules & features in the data.
COMPARISON BETWEEN DATA MINING & DBMS
DATA: Queries (SQL) based on the data held.
DATA MINING: Infer (knowledge required) from the data held to answer queries.
CHARACTERISTICS OF DATA MINING
1) Large quantities of data.
2) Noisy, incomplete data.
3) Complex data structure.
4) Heterogeneous data stored in legacy systems.
Decision Support System – Prof. Chandan Bhattacharya – 27th June, 2006.
DIFFERENCE BETWEEN DATA MINING AND MACHINE LEARNING
Data Mining or Knowledge Discovery in database is about finding understandable knowledge where as Machine Learning is concerned with improving performance of an agent like training a Neural Network.
Data Mining is concerned with very large & real world databases whereas Machine Learning typically looks at smaller data sets.
A data warehouse can be defined as any centralized data repository, which can be required for business benefit.
Warehousing makes it possible to: -
(i) Extract achieved operational data.
(ii) Overcome inconsistencies between different legacy data format.
(iii) Integrate data through put throughout an enterprise regardless of location, format or communication requirements.
(iv) Incorporate additional or expert information.
CHARACTERISTICS OF DATA WAREHOUSE SUBJECT ORIENTED
Data organized by subject instead of application. It contains only the information necessary for decision support process.
1) Integrated – Encoding of data is often inconsistent.
2) Mass user scalability – Access to warehouse to support concurrent users while maintaining acceptable query performance.
3) Network data-warehouse – Data warehouse rarely exist in isolation. Users must be able to look at and work with multiple warehouses from a single client workstation.
4) Warehouse administration – Large scale and time cyclic nature of the data warehouse demands administrative flexibility
5) The RDBMS must integrate dimentioned analysis.
6) Advanced query functionality.
7) End users require advanced analysis calculations sequential comparative analysis to detailed and summarized date.
DATAWAREHOUSING & ONLINE TRANSACTION PROCESSING (OLTP)
Decision Support System – Prof. Chandan Bhattacharya – 4th July, 2006.
DATA WAREHOUSING SYSTEMS
Data Warehouses are interested in query processing as opposed to transaction processing. It contains a place for storing data that are 5-10 years old. These data is used for consequences, trends & forecasting.
1) Preserves the security 7 integrity of mission critical OLTP applications.
2) Give access to the broadest possible base of data.
1) Data is transformed 7 delivered to the data warehouse on a selected model (of mapping definition)
1) Information describing the model & definition of the source data element.
Removal of certain aspect of operational data such as low level transaction information, which slow down the query time.
Processed data transferred to the data warehouse, a large database on a high performance box.
USE OF DATA WAREHOUSE
A central store against which the queries are own, it uses very simple data structures with very little assumptions about the relationships between data.
A data mart is a small warehouse, which provides subsets of the main store, depending on the requirement of a specific group or department. Data marts often use multidimensional databases, which can speed up query processing as they can have data structures, which reflects the most likely queries.
CRITERION FOR A DATA WAREHOUSE
1) Load performance – require incremental loading of new data on a periodic basis.
2) Load processing – data conversion, filtering, reformatting, integrity checks, physical storage, indexing, metadata update.
3) Data quality management – Ensure consistency & referential integrity in massive database size.
4) Query performance – must not be slow b the query of data warehouse RDBMS. It must support modular parallel management.
ARTIFICIAL INTELLIGENCE FAMILY
Artificial Intelligence is commonly defined as the effort to develop computer based systems that behave as human. Such systems would be able to learn natural languages, accomplish coordinated physical task (robotics) & emulate human expertise & decision making (expert system) such systems will also exhibit logic, reasoning, intuition & common sense that is associated with human being.
The figure above illustrates the major branches of artificial intelligence family.
The birth of AI is based on to different approaches
1) Top down
2) Bottom up
Bottom up approach is an effort to build a physical analogy to the human brain while top down approach is the effort to develop analogy to how brain works.
Decision Support System – Prof. Chandan Bhattacharya – 11th July, 2006.
EVOLUTION OF ARTIFICIAL INTELLIGENCE
TOP DOWN APPROACH
One of the first top down efforts was Logic Theory in 1950 applying which a software was developed named Logic Theories that mimicked deductive logic:
Selecting the correct rules and postulates so as to create a coherent logical chain. From these developments emerged expert system, which consist of a limited number of rules for a very specific and limited domain of human expertise.
BOTTOM UP APPROACH
The beginning of contemporary AI started with the concept of feedback to develop a theory of how brain works. According to this theory a brain is composed of millions of neuron cells which processed binary numbers those were connected into a network that took feedback or information from the environment. Learning was simply a matter of teaching the neuron in a brain how to response to the environment.
SYMBOLIC ARTIFICIAL INTELLIGENCE
Symbolic AI (classical AI) is the branch of AI that concerns itself with attempting to explicitly represent human knowledge in a declarative form (i.e. with facts and rules). If such an approach is to be successful in producing human life intelligent, then it is necessary to translate procedure knowledge possessed by human into an explicit form. Game playing programs are the best human experts. One difficult problem encountered by symbolic AI, is Common Sense Knowledge Problem.
STRONG AI AND WEAK AI
Strong AI is the view that a sufficiently programmed computer would actually be intelligent and would think in the same way that a human does where as weak AI is the use of methods modeled on intelligent behaviour to make computers more efficient at solving problems.
HUMAN v/s ARTIFICIAL INTELLIGENCE
1) Human intelligence is a way of reasoning
2) One part of human intelligence can be described as the applications of rules, based on human experience and genetics.
3) Human intelligence is the way of behaving. Even if human do not actually invoke rules, they are obligated to act as if they did by a culture and a society that values reasonable and intelligent behaviour.
4) Human intelligence includes the development and use of metaphors and analogies. Using metaphors and analogy human creates new rules, apply old rules to an unknown/new situation and at times instinctively apply without rules.
5) Human intelligence includes the creation and use of concept. Humans have a unique ability to improve a conceptual apparatus on world around them.
6) ***************************************************************************** Artificial intelligence refers to an effort to develop machines that can reason, behave compare and conceptualise like human being.
How brain works?
(Find out from google.com + howstuffworks.com + wikipedia.org + techtutorials.com )
An expert system is knowledge intensive program that solves a program by capturing the expertise of a human in limited domains of knowledge and experience . An expert system can assist decision making by asking relevant questions and explaining the reasons for adopting certain actions. Some of the common characteristics of expert systems are the following:
1) Perform some of the problem solving work of humans.
2) Represent knowledge informs such as rules or frames.
3) Interact with human.
4) Can consider multiple hypotheses simultaneously.
How expert systems work?
COMPONENT OF EXPERT SYSTEM
4 BASIC ELEMTS OF EXPERT SYSTEM
1) Knowledge base
2) Development Team
3) At Cell
4) The User
This is the modeling of human knowledge that a computer can deal. This model of human knowledge used by expert system is called knowledge base. A standard structured programming construct is the If Then construct in which a condition is evaluated. The difference between a traditional program and a rule base expert system program is the degree and magnitude.
Symantec Nets can be used to represent Knowledge when the knowledge base is composed of easily identified objects of interrelated characteristic. Symmetric net can be much more efficient than rules.
Frames also recognised knowledge into chunks but the relationship is based on shared characteristic rather than hierarchy. This approach is grounded as human use frames a concept to make rapid sense out of perception.
An AI development team is composed of one or several experts who have command over knowledge base and one or more knowledge engineers who can translate the knowledge as described by the expert with a set of rules, frames of Symantec Nets.
A knowledge engineer is similar to a traditional system analyst but has special expertise in collecting information and expertise from other professionals.
The AI shell is the programming environment of an expert system. They are user friendly development environment that can quickly generate user interface screen to capture the knowledge base and manage the strategies of searching the rule base.
The role of the user is both to pose questions of the system and to enter relevant date to guide a system alone. The user may employ the expert system as a source of advice or to perform tedious and routine analysis tasks.
PROBLEMS WITH EXPERT SYSTEMS
1) Expert systems are not applicable for complex managerial problems.
2) Many experts cannot express their knowledge using an IF THEN format.
3) Expertise is collective
4) It may be distributed throughout an organisation
5) Expert systems are extensive to maintain.
6) A more limited role of an expert system.