Chapter 9 Analytics

Generally speaking, we distinguish between three types of forecasts:

(a) short run (up to one year), where the forecast is used mainly in deterministic (certainty) models, b) intermediate term (1-3 years); and (c) long run (beyond three years), where the forecast is used in both deterministic and probabilistic models.

Many DSS development tools include built-in quantitative models

(e.g., financial, statistical) or can easily interface with such models.

If the managerial problem involves multiple goals, one can use the following two-step approach:

1. Select a primary goal whose level is to be maximized or minimized. 2. Transform the other goals into constraints indicating acceptable lower and upper limits, which must only be satisfied. For example, one may attempt to maximize profit (the primary goal) subject to a growth rate of at least 12 percent per year (a secondary goal). There are many alternative approaches to multiple-objective optimization. One is to weight the goals based on importance into a single objective function. Another seeks to obtain the decision maker's economic utility functions for each goal. This is beyond the scope of this material.

Time-series Analysis.

A time series is a set of values of some business or economic variable, measured at successive (usually equal) intervals of time. For example, quarterly sales of a firm make up a time series, as does the population in a city (e.g., counted annually), the weekly demand for hospital beds, and so on.we believe that knowledge of past behavior of the time series might help our understanding of (and therefore our ability to predict) the behavior of the series in the future. In some instances, such as the stock market, this assumption may be unjustified

Association or Causal Methods.

Association or causal methods include data analysis for finding data associations and, if possible, cause-effect relationships. They are more powerful than the time-series methods, but they are also more complex. Their complexity comes from two sources: First, they include more variables, some of which are external to the situation. Second, they use sophisticated statistical techniques for segregating the various types of variables. Causal approaches are most appropriate for intermediate term forecasting.

orecasting models are an integral part of many MSS. One can build a forecasting model or one may use preprogrammed software packages such as

Autobox (Automatic Forecasting Systems, Warminster, PA), and ForecastMaster (Scientific Systems Inc., Woburn, MA), ForecastPlus (Stat Pac Inc., Edina, MN), FUTURCAST (Futurion Inc., Ridgewood, NJ) and SmartForecast (Smart Software Inc., Belmont, MA). Also, many MSS development tools (e.g., financial planning languages and spreadsheets) have some built-in forecasting capabilities.

Decision analysis

Decision situations that involve a finite and usually not too large number of alternatives. the alternatives are listed in a table or a graph, with their forecasted contributions to the goal(s) and the probability of obtaining the contribution.

For optimization of problems with few alternatives and you want to find the best solution from a small number of alternatives you use

Decision tables, decision trees, analytic hierarchy process.

Tools for AHP

Expert choice. Web-HIPRE

For heuristics models and you want to find a good enough solution, using rules you use

Heuristic programming, expert systems

For optimization via an analytic formula and you want to find the best solution in one step, using a formula you use

Inventory models.

Tools for decision trees

Mind tools, tree age software, palisade corp.

DSS modeling (optimization and simulation) contribute to organizational success. examples include

Pillowtext, Fiat, Procter and Gamble.

P=

R-C

For simulation problems, and you want to find a good enough solution or the best among the alternatives checked, using experimentation you use

Several types of simulation.

Two kinds of models

Static analysis (single snapshot of the situation, single internal and steady state) and Dynamic analysis (dynamic models, scenarios that change, time dependent, trends and patterns, more realistic, extends static models)

Trial-and-error sensitivity analysis

The impact of changes in any variable or in several variables can be determined through a simple trial-and-error approach. You change some input data and solve the problem again. When the changes are repeated several times, better and better solutions may be discovered.

two most common methods of sensitivity analysis.

What-if and goal seeking

An influence diagram is

a graphical representation of model; that is, it is a model of a model.

Many decision makers accustomed to slicing and dicing data cubes are now using OLAP systems that

access data warehouses.

Decision tables and decision trees can model

and solve simple decision-making problems.

Sensitivity analysis (as it relates to modeling)

attempts to assess the impact of a change in the input data or parameters on the proposed solution (the result variable). It is extremely important in MSS because it allows flexibility and adaptation to changing conditions and to the requirements of different decision-making situations, provides a better understanding of the model and the decision-making situation it attempts to describe, and permits the manager to input data in order to increase the confidence in the model.

The two types of sensitivity analyses are

automatic and trail-and-error.

3 levels of knowledge

certainty, risk, uncertainty

Decision tables

conveniently organize information and knowledge in a systematic, tabular manner to prepare it for analysis.

Single goal situations can be modeled with

decision tables or decision trees. multiple goal can be modeled with several OTHER techniques.

Linear programming steps

decision variables, objective function, objective function coefficients (ratio between input output) find constraints, formulate model (most time consuming)

Decision variables

describe alternative courses of action.

Analytic hierarchy process (AHP)

developed by thomas salty is an excellent modeling structure for representing multi-criteria (multiple goals, objectives) problems- with sets of criteria and alternatives (choices) commonly making problem into relevant criteria and alternatives. separates the analysis of the criteria from the alternatives, which helps the decision maker to focus on small, manageable portions of the problem. fairly structured.

Although olap systems may make modeling palatable, they also eliminate many important and applicable model classes from consideration, and they

eliminate many important and subtle solution interpretation aspects. modeling involves much more than just data analysis with trend lines and establishing relationships with statistical methods.

To solve a what-if case, using a formulae you use

financial modeling, waiting lines.

For predictive models and you want to predict the future for a given scenario you use

forecasting models, Markov analysis

Coefficients of the Objective Function. T

he coefficients of the variables in the objective function (e.g., 45 and 12 in the blending problem) are called the profit (or cost) coefficients. They express the rate at which the value of the objective function increases or decreases by including in the solution one unit of each of a corresponding decision variable.

Objective Function.

his is a mathematical expression, given as a linear function, that shows the relationship between the decision variables and a single goal (or objective) under consideration. The objective function is a measure of goal attainment. Examples of such goals are total profit, total cost, share of the market, and the like.

Counting Methods.

involve some kind of experimentations or surveys of a sample data with an attempt to generalize about the entire market. These methods are primarily used for forecasting demand for products/services, a part of marketing research. This type of forecasting is quantitative, based on hard data and thus generally considered more objective than the previous types

There is a continuing tend toward making analytics models completely transparent to the decision maker. For example, multidimensional analysis (modeling),

involves data analysis in several dimensions. Data are generally shown in a spreadsheet format, with which most decision makers are familiar.

Mathematical programming

is a family of tools designed to help solve managerial problems in which the decision maker must allocate scarce resources among competing activities to optimize a measurable goal. For example, the distribution of machine time (the resource) among various products (activities) is a typical allocation problem.

Automatic Sensitivity Analysis

is performed in standard quantitative model implementations such as LP. IT reports the range within which a certain input variable or parameter value (uni cost) can vary without having any significant impact on the proposed solution. usually limited to one change at a time, and only for certain variables. establishes ranges and limits quickly.

LP problem characteristics

limited quantity of economic resources, resources are used in the production of products or services, multiple ways (solutions, programs) to use the resources, each activity (product or service) yields a return in terms of the goal, allocation is usually restricted by constraints.

For optimization via algorithm problems, and you want to find the best solution form a large number of alternatives, using a step-by-step improvement process you use

linear and other mathematical programming models, network models

LP is the most common

mathematical programming method. It attempts to find an optimal allocation of limited resources under organizational constraints.

he major use of forecasting, as it relates to modeling, is to predict the value of the

model variables (often demand), as well as the logical relationships of the model, at some time in the future.

Current trends in modeling

model/solution libraries. web-based modeling (optimization/simulation, multidimensional analysis, influence diagrams)

The latest development in improved forecasting techniques is the use of

neural networks (see Chapters 16 and 17), and examples of methods and applications in Hill et al. (1996).

The major parts of an LP model are the

objective function, the decision variables, and the constraints.

Mathematical programming is an important

optimization method.

intermediate result variables

reflect intermediate outcomes in mathematical models. example is employee salary. this constitutes a decision variable for management: It determines employee satisfaction (intermediate outcome) which ultimately determines the productivity level (the final result).

result variables

reflect the level of effectiveness of a system; how well the system performs or attains its goals. the variables are outputs.

Decision tree

shows the relationships of the problem graphically and can handle complex situations in a compact form. but can be cumbersome if there are many alternatives or states of nature. used for scenario analysis.

Analytic hierarchy process (AHP)s a leading method for

solving multicriteria decision-making problems.

Models can be

static (i.e., a single snapshot of a situation) or dynamic (i.e., multiperiod).

Linear Programming (LP)

the best-known technique in a family of optimization tools called mathematical programming. All relationships among the variables are linear. used extensively in DSS. used in supply chain management, product mix decisions, routing, and so on.

Influence diagrams graphically show

the interrelationships of a model. They can be used to enhance the use of spreadsheet technology.

Environmental scanning and analysis

the monitoring, scanning, and interpretation of collected information.0

Sensitivity analysis

the process of assessing the impact of change in inputs on outputs. helps to eliminate variables, revise models to get rid of too large sensitivities, details about sensitive variables or scenarios, obtain better estimates or sensitive variables, reduce sensitivities. can be automatic or trail and error.

In non-quantitative models,

the relationships are symbolic or qualitative.

The most popular end-user modeling tool

the spreadsheet. because it incorporated many powerful financial, statistical, mathematical, and other functions. Spreadsheets can perform model solution tasks such as linear programming, and regression analysis.

Quantitative models are

typically made up of four basic components. result variables, decision variables, uncontrollable variables, and intermediate result variables.

Spreadsheets have many capabilities, including

what-if analysis, goal seeking, programming, database management, optimization, and simulation.