Introduction
The decision tree method is a way that can be used in getting help to make good decision choices, particularly for those decisions which are associated with high risks and costs. This method uses a graphical approach in order to compare various alternatives by assigning specific values to each of the alternatives while considering the uncertainties, payoffs and costs as numerical values. This method is of great help to business analysts, project managers and any level of project decision maker. This method is helpful in a variety of field including but not limited to artificial intelligence, medical diagnosis, game theory, cognitive science, data mining, formal problems solving etc. This method is also used for calculating conditional probabilities in statistics.
The decision tree method is basically a tool to assist decision making process which uses models or graphs like tree and its branches depicting various possible decisions along with their outcomes. It also includes information on costs, utilization, and chances of each decision/outcome. This method has a starting single point generally which spreads out like branches of a tree into various possible alternatives. In this method each alternate may have further alternative options causing the initial branches to have further sub-branches. The overall structure of this method in graphical way, looks similar to a tree. More the alternates and level of alternates are there, more the extension of branches will be .
Advantages of decision trees
There are various methods of analyzing decisions as used by experts. The method of decision tree for decision making has quite enough advantages to make it a valid method to be used by decision makers in various occasions. This method offers following key advantages over other decision making methods :-
• Graphic: The various alternatives for taking decision are depicted on the paper graphically along with their consequences and chances. The schematic diagram used by this method has quite simple approach to address the issue in hand. The visual representation of decision tree is very much helpful especially in comprehending outcome dependencies and sequential decision making.
• Efficient: The various complex alternatives in this method are expressed quickly and clearly. The modification of decision tree is also very easy and simple which may consist of editing, removing and adding any information in the existing decision tree. The effects of editing the input values can easily be seen in the outputs in decision trees. A standard notion of decision tree is very easy for any decision maker to comprehend and utilize in making almost any kind of decision.
• Revealing: In decision tree method, the competitive alternatives can be analyzed and evaluated to some extent without even accounting for the exact values of risks and costs for each alternative. For this way of applying the decision tree method, a term Expected Value (EV) is used which includes relative costs of each alternative, uncertainties in each alternative, and the anticipated payoffs. The Expected Value (EV) reveals the general merits in terms of computing the alternatives.
• Complementary: The method of decision tree for decision making can be used along with other decision making methods/tools which ensure further reliability of the decision in terms of achieving the objectives. This method helps in evaluating the project schedule as well.
Basic Concepts
There are few basic concepts in decision tree method which are needed to be mastered in order to utilize this method in its true essence. These concepts of decision tree method are analyzed in depth in this research paper for the sake of understanding this universal method of decision making. Following are the critical basic concepts needed for an individual to be able to fully understand the method of decision making through the use of decision trees .
1. Nodes and Branches
The notions of decision tree method two basic elements known as nodes and branches. Any kind of decision tree consists of nodes and branches directly proportionate to the number of decision alternatives. There may be multiple possible alternatives and their corresponding outcomes in a decision tree which are drawn by nodes and branches. An outcome in the decision tree may also depend on another outcome within the tree, so in order to show the relationship between both of them a simple node is placed at the end of first outcome in the order of occurrence and the second outcome is places at the branch which is extended from the first outcome. This situation is known as dependent uncertainty. The various decisions in the decision tree may be linked through a sequential connection of branches and nodes. The decision tree notion helps in keeping the connection among various alternatives and outcome simple and understandable.
Root Nodes and Decision Nodes
The small squares in a decision tree identify the decision nodes whereas a decision tree generally begins with certain given initial condition. This initial condition is known as the root node of the decision tree. For instance, in case of medical emergency, the root node will be the representation of the situation of performing an operation, trying chemical treatment, or waiting for any other opinion. The root node is drawn at the very left side of a standard horizontal decision tree.
Chance nodes
The chance nodes in a decision tree are marked by small circles. These nodes represent a specific event which may result in more than one results. In this method two decision alternatives are connected to the chance nodes. The chance nodes might also lead to not only more than one decisions but also to further more than one chance nodes.
Endpoints
A termination or simply known as end point, represents the end of the decision tree or simply speaking a final result for a specific line of branches. A small triangle is placed at the end of the branch to indicate the endpoint of it.
Branches
In a decision tree the lines connecting the nodes are known as branches. Those branches which begin from a decision are known as decision branches whereas those branches which begin from a chance are known as chance branches. A branch may connect itself to any type of node.
2. Payoff values
A payoff value in a decision tree method is equivalent for the net loss/profit expected as a result of any alternative. The payoff values are written at the respective end points of each branch. It is generally in practice to use the financial terms including costs etc. in business applications. The payoff value is basically the difference value of gross revenue to the investment cost. The payoff values might be negative or positive. In case of negative payoff the outcome is considered as loss and in case of positive payoff the outcome is considered as profit.
3. Outcome probability
Any chance node may lead to more than one outcomes in which each outcome is represented by a branch. For instance in a game based on chance, every outcome has a specific probability associated with it for it occurrence. The sum of all the results or outcomes in the given chance nodes is equal to a value of 1 or in term of percentage a 100%. The probabilities are mentioned in parentheses on the chance branches in proper decimal fractions.
4. Expected value (EV)
Expected value or EV is a method to measure relative merits for each decision alternative. The EV term is basically a mathematical combination for both payoffs and probabilities. The EV is calculated after identification of all the payoffs and probabilities. The decision alternative for which the EV value is the highest is the alternative which must be preferred over others. The easiest way of finding out the EV value for each alternative is through calculating EV for every terminated branch and then for every chance node and every decision.
Decision Tree Analysis
The EV is the value at root node represents that a decision of waiting for other reporting methods is better or not. So in case of standard tree analysis the value of EV determines the most suitable alternative for every decision making process. The decision making method employing EV makes it easier and simple to analyze the overall effectiveness and consequences of the alternatives .
Conclusion
Decision tree method is a strong tool for analyzing various alternatives on a single sheet consisting of graphical depiction of the alternatives. This method of decision making has emerged as a preliminary method of evaluating a project as a whole or a part of it. Experts may use this method of decision making along with other tools in order to ensure good decision making.
Works Cited
Barry, De Ville. Decision Tree for Business Intelligence and Data Mining. Cary (NC): SAS Institute, 2006. Print.
Gary, B. Shelly, J. Cashman Thomas and J. Rosenblatt Harry. Systems Analysis and Design. Stamford (CT): Cengage Learning, 2009. Print.
Lior, Rokach and Z. Maimon Oded. Data Mining with Decision Trees: Theroy and Applications. Singapore: World Scientific, 2008. Print.
Shuning, Wu. Optimal Instance Selection for Improved Decision Tree. Michigan: ProQuest, 2007. Print.
