Discovering Information Systems by Jean-Paul Van Belle, Jane Nash, et al - HTML preview

PLEASE NOTE: This is an HTML preview only and some elements such as links or page numbers may be incorrect.
Download the book in PDF, ePub, Kindle for a complete version.

8.3.3 Decision Support Systems

Decision Support Systems (DSS) are systems that assist managers with very specific types of decision-making situations. Though they are often used by the same managers who also rely on MIS, a distinguishing feature is their modelling capability. DSS use various mathematical and statistical models to help the manager generate alternative decision options and evaluate their outcomes. Another difference between DSS and MIS is their time perspective: an MIS

typically produces reports based on historical information where a DSS allows the manager to see the future impact of his decision. Table 8-4 lists some more differences between MIS and DSS.

96

Discovering Information Systems

Section III

8. Business Support Systems

MIS

DSS

Problem type

Mainly for more structured

Good at handling unstructured

problems or programmed

problems.

decision-making.

Support

Provides only the information

Supports all stages in the decision

necessary to make a decision.

making process, including the

modelling and evaluation of various

decision alternatives.

Approach

Typically based on regular,

Excels at interactive and ad-hoc

structured reports.

queries.

System

Often based on printed reports, On-line and real-time interaction,

via batch mode, delivered to

mainly screen oriented.

managers on a regular basis.

Speed

Requires a greater turn-around

Quick

time and is less flexible in

terms of report format

changes.

Development

By IS staff.

Often by end users.

Table 8-4: Some differences between MIS and DSS

When information is required to support management decision making, the user of the DSS is able to interact directly with the computer via a graphical user interface or control language to request the relevant data, select and operate the appropriate decision model and generate the output report in the format required. The following diagram of a DSS shows the three main components; the database, model base and user interface.

Model Base

User

User

Interface

DSS

Database

Figure 8-5: Components of a Decision Support System

DSS Database. This database contains current and historical data from all the relevant business applications. However there are a number of good reasons why organisations do not allow the DSS to access the operational database used by the transaction processing systems, but rather construct another database for this purpose. One key reason for limiting access to the operational database is that DSS requests often require many passes of the database to select the required data. This activity will impact on the service the DBMS can provide to online applications in areas where response times are critical. In addition, while DSS systems normally download a copy of data for analysis and seldom update the database, management is always concerned about the security and integrity of the operational database and prefers to limit access to a minimum. Finally it makes good sense to maintain a database specifically for DSS queries. In the DSS database, some data need be held only in summary form while

certain historical records must be retained for a five year period to allow for trend analysis. In some cases the data may come from different databases, sometimes held on different

Discovering Information Systems

97

8. Business Support Systems

Section III

hardware and software platforms, and the transfer (and potential reformatting) of data to a common DSS database enables queries to be generated combining data from these varied

sources. With the demand for end-user access to current and historical information, many organisations are building data warehouses, which store large quantities of data obtained from different functional areas of the organisation.

Model base. This is a library of analytical tools that can be used to evaluate and represent data. Typical examples are the standard business functions (for example to calculate

discounted cash flows and depreciation), statistical functions (means, standard deviation and variance), data retrieval tools to select, sort and summarise, and the ability to test possible scenarios through sensitivity analysis and goal seeking.

User Interface. In the past managers communicated their report requirements to

programmers who coded the request and delivered the required output. The nature of

management decision making is such that response time is often critical. In addition the solution to problems of this nature is iterative, as one report may trigger the need for alternate investigations into other areas of the business. Often senior managers use a ‘chauffeur’, an analyst or skilled end user to assist them in developing the required DSS output. Obviously the most suitable DSS environment is to have the decision maker interface directly with the DSS. Graphic user interfaces and the increased level of computer literacy within the

management hierarchy, have made this possible.

When a user requires a report or enquiry to be performed by the DSS, he or she will enter the request in a high level, user friendly business language. For example, by clicking the mouse on options and items in list boxes, the user can pick fields, choose selection criteria, detail sequence and request the level of detail for a particular report. The user interface software will then translate this request into the code required to perform the process using the required data and business rules. These user interfaces also offer sophisticated output formatting with, for example, the results being presented in text format or as business charts.

8.3.4 Executive Information Systems

Decisions made by top-level executives are often too unstructured to be adequately supported by a DSS. For this reason, Executive Information Systems (EIS) have been developed, which provide rapid access to both internal and external information, often presented in graphical format, but with the ability to present more detailed underlying data if it is required.

An EIS will continuously monitor selected key performance indicators that have been

identified as critical to the success of the organisation. The user will be alerted to any significant changes that occur, and “drill-down” capabilities will then provide further levels of detail underlying this information. Trend analysis can be done using forecasting models, usually through the integration of the EIS with a DSS system.

8.4 Strategic Systems

An important special type of organisational information system is used to secure or sustain competitive advantage in the market place: strategic information systems. Although these systems generally form part of a more generic business marketing strategy, the information 98

Discovering Information Systems

Section III

8. Business Support Systems

technology is actually a critical enabler or support element to achieving success in the market.

The following three possible strategies are typically distinguished.

Low-cost strategy: use of the information systems to produce significant cost-savings and thus offer services/products at a lower price than competitors (or increase profit margins).

One typical example is the use of alternative marketing or distribution channels such as using the Internet for receiving orders and eliminating middlemen such as wholesalers and retailers. Another example is having more integrated or even completely redesigned

logistics processes such as having a fully automated on-demand production line where

parts and components are supplied on a just-in-time (JIT) basis by the suppliers, often using an Electronic Data Interchange (EDI) system.

A differentiation strategy: use of information systems to provide a distinctive quality or otherwise add value to your products or services. Technology can enhance quality through better manufacturing processes (e.g. a quality control system or automated manufacturing) or add value by increasing the information content or information intensity of the service/product. Other possibilities are the use of computer and communication

technologies to enhance after-sales support e.g. automated (self-)diagnostics, remote

diagnostics, direct internet support and help-line.

A niche marketing strategy: using information systems technology to service very small, isolated or exclusive markets that have specific demands. A good example is the

publishing industry: the ability of the internet to reach individuals with extremely specific interests for whom there would otherwise be no cost-effective marketing or distribution

channel allows the marketing of extremely specialised books. Another example in the

publishing technology is the production of small print runs of “customised textbooks on

demand” whereby course lecturers can compile individually customised textbooks made

up of different modules tailored to their specific curriculum requirements.

8.5 Intelligent Systems

The traditional approach to solving problems using a computer is to provide instructions to the machine (in the form of a program) detailing exactly how the task is to be performed.

However, some tasks are seen as too unstructured (ill-defined) to be programmable. For

example imagine writing a program to control a robot housekeeper. The number of possible situations the robot needs to identify and respond to are so great that conventional

programming techniques are totally inadequate. Add to these problems the concept of

approximate or fuzzy logic. Pure propositional logic (things are either true or false) is too exact for real world problems where humans are required to make guesstimates based on

probabilities.

The ability of computers to work intelligently (and not just follow a set of standard

instructions) has fascinated scientists, researchers and sci-fi writers since the early 1950s, and artificial intelligence (AI) is the branch of computer science concerned with understanding the nature of human intelligence with the goal of simulating aspects of it with a computer.

While we still know very little about how the brain functions, there are four areas of AI research that have made some progress towards the goal of an intelligent machine:

Discovering Information Systems

99

8. Business Support Systems

Section III

Natural languages: the ability for computers to understand the spoken word

Robotics: where machines perform co-ordinated physical tasks

Visual perception: the ability of machines to recognise visually shapes and objects

Expert Systems: systems developed to simulate the decision-making behaviour of humans in a narrow area of expertise.

8.5.1 Expert Systems

Expert systems (often termed knowledge-based systems) are the category of AI which has been used most successfully in building commercial applications. As discussed earlier,

attempting to provide computers with the intelligence to handle complex real world

environments is beyond our current capabilities. However expert systems are knowledge

intensive programs that capture the expertise of a human within a narrow problem domain.

In order to address a particular problem area (for example to diagnose a disease in a sick patient) the expert system must have a knowledge base (a database of facts, intuition and rules about the knowledge domain) and an inference engine (software capable of manipulating

knowledge contained in the knowledge base). Since knowledge takes many forms and shapes

(facts, rules, relationships between facts and rules, probabilities, defaults and exceptions, models, contradictory statements and many other non-structured items), standard database models such as the relational model are not always suitable. Several different knowledge-based databases have been developed for use in expert systems, although despite attempts at standardisation, most of them use their own proprietary ways of representing and dealing with the data.

There are already many commercial examples of expert systems being used across a range of applications such as the diagnosis and treatment of medical conditions, the evaluation of loan applications, and the identification of mineral deposits.

8.6 Data Mining and OLAP

8.6.1 Data warehouses

Business managers have only recently started to realise how much valuable information is hidden inside the many different databases underlying their information systems. Data

warehouses can be used to correlate and analyse the information contained in different

databases within the same organisation. Usually, a copy of the continuously changing

transaction data within the various operational databases is made periodically into one single, huge database: the data warehouse. This data warehouse then contains detailed historical data for all or most of the organisation’s operations. Powerful statistical and charting tools assist managers in comparing and analysing the data.

Effective construction of a data warehouse must provide facilities to integrate data from different functional areas of the business, which may be represented using different formats.

Software programs will extract data from its original source, convert it into a uniform format and then store it in the warehouse.

100

Discovering Information Systems

index-113_1.png

Section III

8. Business Support Systems

Sometimes it is technically not feasible or it is too expensive to merge the data from many databases. In those cases, some benefit can already be had from focussing on the

organisation’s most important databases, usually the ones relating to customers and sales.

Such a mini data warehouse focussed on one particular functional area is called a data mart.

8.6.2 On-line analytical processing

Imagine a marketing manager trying to analyse this quarter’s sales to discern trends and pinpoint problem areas. How can the data be analysed? There are many possible views.

• According to different product groups.

• Using the time dimension, comparing the different months.

• Looking at the cost components and contribution to income.

• Comparing sales budgets, projections, actual sales and variances.

• Analysing sales according to geographic region.

• Evaluating the various marketing channels: wholesales, sales force, brokers and the

corporate market.

On-line analytical processing (OLAP) is concerned with the real-time analysis of large corporate databases to find trends and inter-relationship by managers and decision-makers.

The user formulates complex queries and searches by means of sophisticated, interactive

front-end applications such as statistical packages, spreadsheets or decision support systems.

Because traditional database models are not very good at handling and displaying many

different dimensions of business data for simultaneous analysis, multidimensional analysis technologies are used to filter and aggregate subsets of the data. Advanced statistical analysis tools and graphical interfaces are incorporated to facilitate data visualisation, often using hypercubes to display multidimensional information.

Figure 8-6. Example of a hypercube (www.ssa-lawtech.com/ wp/wp2-3.htm)

Discovering Information Systems

101

8. Business Support Systems

Section III

The raw data is often derived from traditional application databases, but because of the huge amounts of data processing involved, a separate server may be used to store and process the data subsets being used for analysis. This approach gives rise to a three-tier model, which includes the server or mainframe which hosts the organisational database, the workstation on the manager’s desk through which queries are submitted, plus a dedicated OLAP server

holding relevant subsets of data in multidimensional databases. Unlike an executive

information system, this does not usually provide the ability to “drill down” to the level of the original transactions underlying a query result.

8.6.3 Data mining

The huge amount of information and the many possible different perspectives often make it very difficult for humans to discern meaningful trends within the masses of data in the data warehouse. Statisticians have developed a number of methods that automate the discovery of non-random trends and significant inter-relations. Data mining is the use of these statistical methods, packaged in a single computer package and let loose on their own, without human intervention, to discover deep or hidden data interrelationships.

8.7 South African Perspective

An example of a locally developed system that supports business transactions, management control and decision-making is Digitot, a computerised beverage-dispensing system.

Traditionally, the hospitality industry loses between 10% and 25% of revenue on alcoholic drinks, due to spillage, pilferage and “freebies”, and frequent manual stocktakes are

necessary. The Digitot system, which is integrated with a point-of-sale system, uses electronic tot measures to digitally dispense and count each tot and automatically update stock levels.

Flexible reporting over selected periods can analyse this data to determine when stock has gone missing and who was responsible for the loss. Data analysis can also reveal patterns in customer preferences and identify products that offer value for money, which can be used as the basis for promotional marketing.

8.8 Beyond the Basics

Neural networks are proving increasingly valuable in complex decision-making, where a large number of factors must be simultaneously considered and human inconsistency or prejudice could affect the decision outcome. Furthermore, they are able to adapt to changes in the business environment over time. The neural network consists of many cooperative processing elements, that are “trained” using a large number of historical transactions (the data that was evaluated and the outcome that was reached) in order to establish criteria that can be applied to future decisions. The weighting of these criteria will subsequently change over time, since each new application of the neural network may affect the previous relationship between

elements.

Neural networks are particularly useful when processing complex data for which no

established decision rule is known, or when there are too many variables to consider. In this case, it is easier to let the network learn from examples.

Neural networks are being used:

102

Discovering Information Systems

Section III

8. Business Support Systems

• in investment analysis: to attempt to predict the movement of stocks currencies etc., from previous data. There, they are replacing earlier simpler linear models.

• in monitoring: neural networks have been used to monitor the state of aircraft engines.

By monitoring vibration levels and sound, early warning of engine problems can be

given.

• in marketing: neural networks have been used to improve marketing mailshots. One

technique is to run a test mailshot, and look at the pattern of returns from this. The

idea is to find a predictive mapping from the data known about the clients to how they

have responded. This mapping is then used to direct further mailshots.

8.9 Exercises

8.9.1 Processing methods

Compare a batch processing system (such as might be found in a hardware store) with a

realtime processing system (e.g. in a bank), in terms of the authorisation methods used to check the identity of the customer, how the initial transaction details are captured, the completeness of transaction data, opportunities for human error to enter the process,

availability of data for managerial decision-making.

8.9.2 Organisational information flows

Construct a diagram to show how data and information flow between the different types of organisational information systems (TPS, MIS, DSS, EIS), and their users, data stores and environment.

8.9.3 Fuzzy logic

Fuzzy logic refers to a type of expert system which caters for non-binary decision-making (i.e.

not simply a yes/no answer). For example, when you buy a car, you might be satisfied (in varying degrees) with petrol consumption falling within a range, rather than arbitrarily selecting a cutoff point below which it is okay and above which it is unacceptable. Search on the Internet or in an IS textbook to find a simple explanation of the concept of fuzzy logic, and an example of how it has been implemented in commercial devices.

Discovering Information Systems

103

9. E-Commerce

Section III

9. E-Commerce

Business was quick to grasp the marketing and business potential offered by the Internet.

Initially, businesses used the Internet to facilitate communication by means of e-mail. This was quickly followed by tapping the web’s potential for the dissemination of product and other marketing information. The provision of advertising space ( banners) on frequently visited web sites is the main source of income for search engines (sites allowing you to search the Internet for information) and web portals (web sites that provide additional value-added personal services such as news, financial information, weather forecasts, items of interest etc.) A number of specialised companies have realised that the Internet can be a direct and

extremely cost-effective channel of distribution. Some companies already have a physical infrastructure and use the web to enhance their distribution channel e.g. you can now order your pizza, bank statements or movie tickets via the web. Other, virtual companies have almost no physical infrastructure and are mere “conductors” for the flow of products of

services.

Important categories of e-commerce include:

• Business-to-consumer (B2C) in which organisations provide information online to

customers, who can in turn place orders and make payments via the internet

• Business-to-business (B2B) in which business partners collaborate electronically

• Consumer-to-consumer (C2C) in which individuals sell products or services directly

to other individuals.

The technologies that are needed to support electronic commerce include the network

infrastructure (Internet, intranets, extranets), software tools for web site development and maintenance, secure ordering and payment methods, and resources for information sharing, communication and collaboration. When e-commerce is done in a wireless environment, such as through the use of cellphones, this is referred to as mobile commerce (m-commerce).

9.1 B2C e-Commerce

Electronic retailing is similar in principle to home shopping from catalogues, but offers a wider variety of products and services, often at lower prices. Search engines make it easy to locate and compare competitor’s products from one convenient location and without being

restricted to usual shopping hours. Electronic malls provide access to a number of individual shops from one website. On-line auctions have also proved a popular way of disposing of

items that need a quick sale.

Business-to-consumer commerce allows customers to make enquiries about products, place

orders, pay accounts, and obtain service support via the Internet. Since customers can enter transactions at any time of the day or night, and from any geographical location, this can be a powerful tool for expanding the customer base of a business. However, the existence of a website does not guarantee that customers will use it, or that they will return to it after a first visit. Firms investing in electronic commerce need to consider a number of factors in

developing and maintaining their e-commerce sites.

104

Discovering Information Systems

Section III

9. E-Commerce

A successful web site should be attractive to look at and easy to use. In addition, it should offer its customers good performance, efficient service, personalisation, incentives to

purchase and security. Inadequate server power and communications capacity may cause

customers to become frustrated when browsing or selecting products.

Many