Metodologia de Modelagem para Suporte à Decisão - Guia Introdutório

Basics of Modeling Methodology A document to accompany the course Modeling to Support DecisionMaking by Prof Dr Georgiou This document introduces the basics of model design and construction that are taught in this course It follows and further develops the leading internationalstandard methodology developed by Wayne Winston and Christian Albright of the Kelley School of Business at Indiana University The issues included in this document are illustrated in the numerous Sample Models that may be found in MSDM Online ArchiveWorkshops on Models for DecisionMakingSample Models Contents General Issues 2 Descriptive Cell Labels 4 Inputs 5 Decisions 6 Constraints 7 The Objective 8 Cells Containing Formulas 9 Graphs 10 General Issues In general a model begins with a title in the top left cell A1 and is designed vertically downwards with sections for Inputs Decisions Constraints Accounting in that order For example notice the placement of the title and section titles in the following model In some models the Decisions and Constraints sections are combined as illustrated below As modeling methodology is mastered a certain flexibility in model design organically emerges The vertical topdown design however is the default template that should be followed as much as possible Descriptive Cell Labels All cells with values whether typed or computed have descriptive cell labels to their left For example The above example also illustrates that all cells concerning money are formatted accordingly The use of cell labels may also be appreciated in the illustrations to be found in the section General Issues of this document Inputs Inputs are of two types i loose inputs and ii participatory inputs In either case an input is typed once and enclosed in a blue border for easy identification For example N ote The red triangle in the input cell above indicates an optional comment has been included Microsoft Excel has a useful comments facility that enhances model communication to users Once an input is typed it is never changed nor is it typed anywhere else in the model However it may be addressed anywhere in the model especially when it is used in computational formulas Loose inputs are placed in a dedicated section at the top left of a model Examples are evident in the illustrations to be found in the section General Issues of this document Inputs may be grouped together if they concern similar concepts for example different available prices in which case the group not the individual cells are enclosed in a blue border For example Participatory inputs are socalled because they are modeled as part of that is participate in constraints See the section Constraints in this document for information Decisions Models are designed to facilitate decisions such as a decision concerning the amount of items to be ordered or one concerned with the spread of capital across a portfolio of investment options For ease of identification decisions are enclosed in a red border For example consider the decision of ordering a certain amount from a supplier Alternatively the decision may concern the amount produced of a certain product Notice from the above illustration that a model may include multiple decisions Decisions are easily identifiable in the illustrations to be found in the section General Issues of this document Constraints Constraints are modeled transparently in the spreadsheet Constraint inequalities are included in the model in the interest of communicative transparency In a constraint t he lefthand side LHS is usually a formula t he righthand side RHS is usually an input Participatory inputs are modeled directly at the constraints such as in the following example The above example illustrates the modeling of constraints in what may be termed the vertical manner because the symbols of the constraints are modeled vertically in a column with the LHS and RHS accordingly placed Constraints can also be modeled horizontally such as in the following example In the above example the top line of the constraint in this case consisting of decisions red cells is known as the LHS and the bottom line of the constraint in this case consisting of inputs is known as the RHS Examples of constraint modeling may also be appreciated in the illustrations to be found in the section General Issues of this document Notice especially how constraints are grouped when they concern common constraint symbols an issue of primary importance in modeling methodology with especial relevance to model layout and the inclusion of Solver parameters discussed in PMS p 80 Also consult the multiple examples of models in the MSDM Online Archive The Objective Models are usually designed to explore an objective such as the amount of possible profit given the parameters of a situation or the amount it will cost to undertake a particular project Objectives such as these are placed in what may be termed target cells and are enclosed in a doublelined black border For example or A model only ever has one target cell It is usually placed at the end of a model as may be appreciated in the illustrations to be found in the section General Issues of this document Cells Containing Formulas Cells containing formulas are not enclosed in colorcoding For example a computation of units sold may be modeled as follows where the value 2000 results from a formula performing a computation Another example is as follows Graphs Graphs are best kept simple and placed alongside the table of data that they serve to illustrate For example Notice that the illustration above demonstrates a key to communicative transparency in graph design A graph illustrates how a variable in the yaxis changes as a function of the variable in the xaxis Therefore the title of a graph is best written as Yaxis title as a function of Xaxis title Also notice all the other elements of this graph design and how they draw their information from the table of data Indeed excellence in graph design is achieved when all of its illustrated information is directly sourced from a table of data such that if the data in the table changes the graph changes automatically Basics of Modeling Methodology Prof Dr Georgiou 2 2