Week 4

Systems and Information Quality

Decision Support Systems What is a System?

A group of interacting components with a purpose

Group – must consist of more than one item Interacting – the components must operate

in some relationship to each other Components – may be elementary items or

other systems Purpose – systems must have a purpose

Systems

Systems have a boundary that separates the components of the system from its environment

Objects or information can cross the boundary in an open system

Nothing crosses the boundary in a closed system

Open and Closed Systems

Closed System Open System

Closed Systems

Are rare in the real world Sometimes useful to treat system

as closed to simplify analysis Example: Auto testing on a closed

track Systems can be of many types

Body Metro Area Transportation System

Feedback

Is output from a system component can be internal of external

Example: Elevator Closed Loop Systems – use

feedback to adjust outputs Open Loop Systems – those that do

not use feedback to adjust outputs

Information System

Is a system whose purpose is to store, process, and communicate information

IS designer’s job is to determine which parts of an info processing task are to be done by a computer and human

DSS as Information Systems

Unique characteristics Uses 1 or more data stores Does not update the data stores Communicates with a decision maker Decision maker supplies DSS specific

information defining the problem

Questions to Start Designing a DSS

What is the purpose? What external entities will the DSS

communicate with? What internal data files does the

DSS use? What are the major processes in

the DSS?

Information

Is anything that reduces uncertainty Example:

The stock gained $5 today vs. The stock went up today

Information is data in its business context

Creating Information

Comparing one data element to another

Performing calculations on data

Information Quality Quality level – degree of excellence

in the context of its intended use High quality info enables users to

make good decisions quickly Low quality info leads to poor

decisions and wastes decision makers’ time

Decisions are no better than the information that they are based upon

Information Quality Factors Relevance Correctness Accuracy Precision Completeness Timeliness Usability Accessibility Consistency Conformity to expectations

Relevance Relevant if it applies to the task

being performed Degree of Relevance Not an all or nothing situation

experience can screen out the unnecessary parts

Data requirements study should be one of the first items in a DSS project

Correctness, Accuracy & Precision

Correctness – based on the right part of the ‘real world’

Accuracy – a measure of how close it is to the ‘real world’ value

Precision – is the maximum accuracy

Completeness

It includes all the necessary elements for the decision that is to be made

Can be traded off for better timeliness or lower cost

Timeliness

Information must be available in time for its intended use

Information must reflect up-to-date data

Usability

How quickly and easily intended users can figure out what they need to know from what they see

Info that forces the use of extra effort to interpret is not very usable

Accessibility

Info can be obtained quickly, with an acceptable level of effort, from anyplace a user is expected to be when they need the information

Consistency

All data elements that contribute to an information item are based on the same assumptions, definitions, time period, and other factors

Conformity to Expectations

How well the information’s processing and timeliness will match what the user expects

Types of DSS File drawer systems allow immediate

access to data items Data analysis systems allow the

manipulation of data by operators Analysis information systems provide

access to a series of databases and small models

Accounting models calculate the consequences of planned actions on the basis of accounting definitions

Types of DSS Representational models estimate the

consequences of actions on the basis of models that are partially nondefinitional

Optimization systems provide guidelines for action by generating the optimal solution consistent with a set of constraints

Suggestion systems perform analysis using user input leading to a specific suggested decision

File drawer systems Provide access to data items

Examples include real-time equipment monitoring, inventory reorder and monitoring systems. Simple query and reporting tools that access OLTP or a data mart fall into this category.

They are the simplest type of DSS Can provide access to data items Data is used to make a decision ATM Machine Use the balance to make transfer of funds

decisions

Data analysis systems Support the manipulation of data by

computerized tools tailored to a specific task and setting or by more general tools and operators

Examples include budget analysis and variance monitoring and analysis of investment opportunities. Most data warehouse applications would be categorized as data analysis systems.

Provide access to data Allows data manipulation capabilities Airline Reservation system No more seats available Provide alternative flights you can use Use the info to make flight plans

Analysis information systems Provide access to a series of decision-oriented

databases and small models Examples include sales forecasting based on a

marketing database, competitor analyses, product planning and analysis. Online Analytical Processing (OLAP) and Business Intelligence (BI) systems generally are in this category.

Information from several files are combined Some of these files may be external We have a true “data base” The information from one file, table, can be

combined with information from other files to answer a specific query.

Accounting and financial model-based DSS Calculate the consequences of possible actions

Examples include estimating profitability of a new product; analysis of operational plans using a goal-seeking capability, break-even analysis, and generating estimates of income statements and balance sheets. These types of models should be used with "What if?" or sensitivity analysis.

Use internal accounting data Provide accounting modeling capabilities Can not handle uncertainty Use Bill of Material Calculate production cost Make pricing decisions

Representational model-based DSS Estimate the consequences of actions on the basis

of simulation models that include relationships that are causal as well as accounting definitions.

Examples include a market response model, risk analysis models, and equipment and production simulations.

Can incorporate uncertainty Uses models to solve decision problem using

forecasts Can be used to augment the capabilities of

Accounting models Use the demand data to forecast next years

demand Use the results to make inventory decisions

Optimization model-based DSS Provide an optimal solution consistent with a

series of constraints that can guide decision making.

Examples include scheduling systems, resource allocation, and material usage optimization.

Used to estimate the effects of different decision alternative

Based on optimization models Can incorporate uncertainty Assign sales force to territory Provide the best assignment schedule

Suggestion DSS Based on logic models that perform the logical

processing leading to a specific suggested decision for a fairly structured or well-understood task.

Examples include insurance renewal rate calculation, an optimal bond-bidding model, a log cutting DSS, and credit scoring.

A descriptive model used to suggest to the decision maker the best action

A prescriptive model used to suggest to the decision maker the best action

May incorporate an Expert System Use the system to recommend a decision Ex: Applicant applies for personal loan

Application to Revenue Management Airlines spend a great deal of effort

on deciding what conditions to apply to each discount fare, how large a discount, how many seats

2 fares Full Fare available until takeoff and

refundable Supersaver 30 day advance purchase,

Saturday night stay, non-refundable

Application to Revenue Management

Dilemma: not enough full fare travelers to fill plane but it can fill with SS fare travelers if the price is right but then no seats left for FF travelers

Result: less profit, customer loss To avoid this the airline must limit the

SS fares to say 40 out of 100 seats

Application to Revenue Management Decision: How many SS fares to sell

If too high turn away FF travelers If too low empty seats

Suppose FF sales normally distributed with a mean of 40.5 FF tickets (decision)

Answer depends on the price ratio If 2:1 ratio then sell 59 or 60 SS tickets because expected revenue

the same If SS fare 2/3 of FF ticket then they would want a 2/3 probability of

filling the FF seats, the 2/3 point is 0.43 standard deviations from the mean so if standard deviation is 14 then offer 65 or 66 SS tickets

Result: increased probability that FF will not find seat from 50% to 67% this is acceptable

If SS fare 1/3 of FF ticket then 53 or 54 SS tickets Result: company accepts higher probability of empty seats but OK

since SS sale less desirable Correct decision: empty seat if probability of FF traveler times FF

price > SS price

Application to Revenue Management

Revenue managers can manipulate 3 variables Number of seats at each price Price Restrictions

Application to Revenue Management

File drawer systems Could find out how many seats were

sold on a given flight at a given fare with a given set of restrictions. Using this info could manually set/modify prices, conditions, limits on future flights

Application to Revenue Management

Data analysis systems Could simplify extracting the

historical data from a database and could calculate averages, trends and similar aggregates from the raw data

Application to Revenue Management

Analysis information systems Could present historical average

booking data for a given type of ticket at a given fare in the form of a graph

Application to Revenue Management

Accounting model Could calculate the expected revenue

for a given seat allocation and the corresponding profit

Application to Revenue Management

Representational model Could start to consider human behavior

such as the price-demand elasticity curve

By combining with accounting models could estimate revenue that will accrue from any mix of restrictions, prices and capacity limits

Application to Revenue Management

Optimization system Since the variables in the airlines

control are few (3) and known, it is easy to program a computer to evaluate them over a range of interest and return the best answer

Application to Revenue Management

Suggestion system Could cope with issues that arise

when several flight segments interact by processing large amounts of data quickly

Your Turn

Questions / Comments / Criticisms