Decision Support Systems (DSS) Information is an organization’s core asset. Transaction processing systems capture a large amount of data. Management information systems and decision support systems process and refine that data to provide vital information to decision makers and problem solvers. As organizations reengineer business processes, information systems designed for decision-making are no longer used solely by managers. To empower employees to make their own decisions and solve their own problems, organizations must provide employees at all levels of the enterprise with timely information provided by these systems.
Decision Making and Problem Solving Problem solving is a critical activity for any business organization. Once a problem has been identified, the problem-solving process begins with decision making. A well-known model developed by Herbert Simon divides the decision-making phase of the problem-solving process into three stages: intelligence, design, and choice. This model was later incorporated by George Huber into the following expanded model – see next slide.
How Decision Making Relates to Problem Solving
1st stage: Intelligence Potential problems or opportunities must be identified and defined. This is the most important step because if the wrong problem is identified and defined, the entire effort of problem solving is wasted. Symptoms are not problems. To distinguish symptoms and real problems, we need to gather data describing the problem.
Gather data about the problem Environmental resources and constraints are investigated during the 1st stage (Intelligence) to gain understanding of the problem. Study the environment which may include suppliers, customers and competitors (from market research) etc. Competitors: sell at prices 10% to 15% lower. Suppliers: increase costs of goods sold to 30% or more because of recent oil price hike. Customers: complain about product’s defects which affect human’s health seriously.
Decision Support System (DSS) A DSS may be expressed as a mathematical program, with some symbolic representations to represent real world objects, quantities and meanings, to solve business decision problems, e.g. shared resource allocation problems (see Tutorial D or case 18 of “Advanced Cases in MIS”).
Shared resource allocation problems They have resource allocation limits which can be expressed using mathematical constraints. e.g. maximum available cotton (C) in stock for T-shirt production in a factory is 10 tons. This limit can be expressed as a constraint: C <= 10 tons. Some constraints may conflict with another, and thus finding solutions or the best solution is a difficult mathematical problem called optimization or linear programming problem.
Programmed Decisions Easy to computerize using rules, procedures, quantitative methods or mathematical formulae. E.g. Simple financial model: Profit = Revenue - Cost Simple Present-value cash flow model: P = F / (1+i)n where P = present value, F = future single payment ($), i = interest rate, n = number of years. Problems/decisions solved by operational or transactional processing systems (TPS) are easily programmed, they are structured problems. Routine, summary reports produced by MIS are also structured problems. When selecting an alternative in the choice stage, various factors affect the decision. We saw in the airport transportation example that resource constraints, such as time, money, or availability, are factors. Another factor is whether the decision can be programmed. Programmed decisions are made by following rules, procedures or quantitative methods that can be described in advance and regularly used, since the situations are recurring and well-structured. Management information systems are designed to provide information to address programmed decisions. Many simple programmed decisions can be completely automated – for example, inventory control systems can be programmed with reorder points and automatically trigger an order for more merchandise when the reorder point is reached.
Nonprogrammed Decisions Rules and relationships not well defined Problem is not routine, exceptional cases Not easily quantifiable Determining an appropriate training program for new employees is an example of unstructured problem. E.g. interviewing new employees is also a nonprogrammed or unstructured decision.
Optimization (using Solver) finds the best solution out of many combinations of possibilities E.g. find the appropriate number of products a factory should produce to meet a profit goal, given certain constraints and assumptions. E.g. E.g. of a problem constraint: there is a limit on the number of working hours per day (X) for each machine in the factory: X <= 8 hours. E.g. minimum number of basketballs to produce in a factory per month (Y): Y >= 40000 balls Computerized decision support systems can usually be used for both optimization and satisficing modeling. An optimization model finds the best solution in relation to the constraints, assumptions, and goals it was given. For example, an optimization model can find the optimal labor cost to produce a particular product and meet a specific level of profit, subject to the cost of raw materials and machinery. Profit level is a goal and costs are a constraint in the model.
Heuristics They are “rules of thumb”, guesses or estimates based on vast experience, or commonly accepted guidelines. E.g. when the inventory level for a certain item drops below 20 units, an experienced manager would order 4 month’s supply as a good guess to avoid out of stock without too much excess inventory. Heuristics are used in optimization for efficiency especially when there are many complicated problem constraints, changing cells and decision variables. Heuristics, or rules of thumb, are often used in decision making. Heuristics are generally accepted guidelines, or guidelines developed through experience, that usually find a good solution. For example, you might follow a heuristic of taking an umbrella if it is cloudy, windy, and humid when you leave the house. Your experience has shown you that generally this results in having an umbrella when it rains. However, this isn’t an optimal solution – since sometimes you carry an umbrella unnecessarily and sometimes it rains on days when you don’t have an umbrella. But the cost of finding an optimal solution is far too great in terms of time and money.
Excel Solver’s optimizing routines
DSS: What-if (sensitivity) analysis Makes several forecasts for different possible economic situations (or scenarios) which could be good, bad or stable by using varying estimates of inputs such as growth rate, oil price, labour change, salaries etc. The varying estimates of inputs reflect uncertainties of real economy. In Supply/Demand, elasticity can be done using what-if analysis.
DSS: What-if (sensitivity) analysis Sensitivity analysis answers the question, "if these variables deviate from expectations, what will the effect be (on the business, model, system, or whatever is being analyzed), and which variables are causing the largest deviations?“ What-if (Sensitivity) analysis is the study of how the uncertainty in the output of a mathematical system can be apportioned to different sources of uncertainty in its inputs.
Decision Support Systems (DSS) A CBIS that focuses on decision-making effectiveness for various decision-making levels: mostly for mid and top levels of management, less for bottom level management. Decisions support systems are people, procedures, software, databases, and devices that are used in problem-specific decision-making and problem-solving. Decision support systems are particularly useful when dealing with semi-structured, poorly structured, or unstructured situations. Although decision support systems are used most often at higher levels of management, all employees may use them to assist in even programmable decisions.
Characteristics of DSS Predictive nature – output information is for future events rather than descriptive of past events, should help reduce risks in future. E.g. forecasts of future economic conditions, projections of new product sales, forecasts of changing target customer groups. Summary form – output information is not detailed, but concerned with global data. E.g. managers are not interested in the details of customer’s invoices but more interested in the overall buying trend in the summaries of sales groups.
Characteristics of DSS Ad hoc basis – strategic planning information is produced irregularly but with a specific purpose. E.g. Managers may request marketing analysis information about a new set of stores (or a new product) when they are considering adding new stores in the region. Unexpected information – economic forecast for the economy and for the industry may often find surprises to managers. E.g. marketing survey in above may produce store locations that had not been expected.
Characteristics of DSS External data – most of the inputs into DSS are from sources external to the firm. Information such as investment opportunities, rates of borrowed capital, census data, economic conditions must be obtained from databases outside the firm e.g. government databases. Subjectivity - input data into DSS are usually highly subjective (personal opinions based on experiences) and their accuracy may be a suspect. E.g. rumors about future stock market trends reported by brokers.