IS Week 10

This week in class we discussed and studied decision trees and their application in dealing with uncertainty in decision making. Decision trees are useful because the analysis of complex decisions with significant uncertainty can be confusing for a variety of reasons. First, and most importantly, the consequence that will result from selecting any specified decision alternative cannot be predicted with certainty. Not only this, but there are also often a large number of different factors that must be taken into account when making a decision. Finally, a decision maker's attitude toward risk taking can impact the relative desirability of different alternatives. Decision tree diagrams are read from left to right. The leftmost node in a decision tree is called the root node. The branches emanating to the right from a decision node represent the decisions that can be made from that node. Only one of these alternatives can be selected. The small circles in the tree are called chance nodes. The number shown in parentheses on each branch of a chance node is the probability that the outcome shown on that branch will occur at the chance node. The right end of branch through the tree is called an endpoint, and each endpoint represents the final outcome of following a path from the root node of the decision tree to that endpoint. A critical component of the analysis of decision trees is the use of expected value. This allows for sequential decision making, once you know what can be expected from each branch of a decision tree. With this more accurate understanding of each decision branch, decision making can be optimized.

https://towardsdatascience.com/decision-trees-in-machine-learning-641b9c4e8052


Comments

Post a Comment

Popular posts from this blog

IS Week 12

This week we continued our discussion of Information Based Processes and Process Analytics. Specifically, we discussed Process Time Analysis. Process Time Analysis is the calculation of the execution time and cost of a process, using expected cost to account for branching. This must first be done via a time study. Time studies measure how long each activity in a process takes, how long a unit has to wait before processing starts, and then sum the processing and wait time to get flow time. Time studies take branching into account by identifying the likelihood of each branch occurring. Execution time is generally estimated by either observation or random sampling. Alternatively, several experts can be called in to give estimates, followed by a delphi analysis that will weight each of these estimates and provide a master estimate. This is known as single band delphi. Alternatively, wide band delphi can be employed. This is where each expert provides a pessimistic, optimistic, and expected...
Read more

IS Week 11

This week in class we covered information based processes, information flows, and process analytics. There are two prominent ways to model information flows. Every information based process has a couple of inalienable traits: an explicit and measurable goal, specific inputs in the form of information, specific outputs in the form of new information, the use of resources, and having an effect on multiple organizational units. Data is generally modeled via Data Flow Diagrams, whereas activities are generally modeled through Flow Charts. The first thing an analyst must do when modeling a process is identify where the process begins and where it ends. If it is ever unclear how detailed a process should be, an analyst should air on the side of generality, choosing instead to refer to subprocesses that can in turn be modeled on their own. In the modeling of processes, different actors become involved. Popular actors include customers, employees, suppliers, and regulatory agencies. Process mo...
Read more

IS Week 2

This week in IS2000 we largely discussed ontologies. An ontology is the design of formal conceptual structures for a given domain of inquiry. The objective of ontologies, and building them, is to represent abstract, real world physical concepts and their relations in a tangible manner. OOIs represent ontologies within and with regard to information systems. Ontologies describe entities, their attributes, and the relationships between them. Entities are things, physical objects, and abstract concepts. Attributes are properties of entities, as well as measurements of quantitative data with regard to them. Examples of relationship matrices include taxonomies, partonomies and associations. Multiplicities and cardinalities are both examples of constraints. We also sought to answer the question of why we use ontologies. The creation of ontologies forces rigorous analysis and facilitates unambiguous discourse about domains. This is critical for information science, and within the field it i...
Read more