Topic: Education Student loans What type of schools have undergraduate students that need to borrow the most amount of subsidized and unsubsidized loans?
Milestone One: Choose a Data Set and Formulate Decision Analysis Research Question
In Milestone One, you are to provide an abstract describing your decision question and high level approach. Consider this abstract as the information that your peers will read and determine if they want to read more. It needs to include a statement of your decision question, what your premise is, and how the data will be used to support the analysis.
The information that follows are additional notes for Milestone One:
In terms of data, the data will lend itself to a decision-making scenario with discrete outcomes. As additional information on how to select your data, first think about the course materials and what you have learned so far about the types of situations that lend themselves to decisions under uncertainty. Then, use the table of student projects as a seed to see what other people have worked on in the past.
Selecting your data set is step one, but it also overlaps step two, which is formulating a research question. Decision analysis research questions are best stated as a discrete set of choices to be analyzed. A good decision analysis research question also possesses the following hallmarks:
· It has clarity of purpose by being framed as a discrete set of choices to be analyzed.
· It is concise.
· It is appropriate and can be answered with decision analysis techniques.
· Its parts are relevant to each other.
· It has not been answered before, or the variation from existing works is great enough to make it a novel line of inquiry.
· It is open-ended enough that it may lead to new research questions.
· But it is closed enough that direct answers are possible, even if they may not be found by this round of analysis.
Overview
You must complete a decision analysis research project as your final project for this course. Your research project will focus on a real-world topic of your choice, as approved by your instructor.
You will pick a topic from the list provided or with approval from your instructor, and create a data analysis plan and decision tree model based on a real-world scenario.
This assessment wi ll provide you with the opportunity to employ highly valued decision support skills and concepts for data within a real-world context. You can use the Final Project Notes document, found in the Assignment Guidelines and Rubrics section of the course.
The project is divided into three milestones, which will be submitted at various points throughout the course to scaffold learning and ensure quality final submissions. These milestones will be submitted in Modules Two, Five, and Seven. The final submission will occur in Module Nine.
This project will address the following course outcomes:
Appraise data in context according to industry-standard methods and techniques for its utility in supporting decision making
Determine suitable data manipulation and modeling methods for decision support
Articulate data frameworks for organizational decision support by applying data manipulation, modeling, and management concepts
Evaluate the ethical issues surrounding organizational use of decision-oriented data based on industry standards and one's personal ethical criteria
Create and assess the agility of solutions through application of data-mining procedures for decision support in various industries
Prompt
Your decision analysis model and report should answer the following prompt: How does your model and evaluation resolve uncertainty in making a decision? In order to produce your analytic report, you will need to choose and investigate a data set using the decision analysis techniques you learned in class. Then you will formulate a research question, write an analytic plan, and implement it.
Your report should not solely consist of descriptions of what you did. It should also contain detailed explorations into the meaning behind your model and the implications of its results. You will also be testing your model's fitness and evaluating its strengths and weaknesses.
The project in a nutshell:
1. Choose a data set (get ideas from the source list in the spreadsheet Final Project Topics and Sources.xls)
2. Formulate your decision analysis research question
3. Write an analytic plan
4. Perform the top-down or bottom-up modeling
5. Perform model diagnostics
6. Evaluate
These activities are broken up into milestones so that the work is spread throughout the term and you can get early assistance with any obstacles.
A decision analysis report is similar to any other analytic report. These reports introduce a problem, state a line of inquiry, explain a model that the author developed, discuss results and limitations, and then make conclusions and recommendations. Some decision models seek the best expected value among a discrete set of choices.
Other decision analyses might seek the threshold values at which the model changes from one recommendation to another, describe the implications, and leave it to the reader to decide what to do. Still other decision models might look for the likeliest path to explain pat terns that are already present in a data set. In all cases, they have something in common: They are trying to help resolve uncertainty. Your job is to bring clarity to the decision being made.
Decision analysis seeks less to produce a definitive result, and more to accurately explain the combinations of possibilities that can lead decision makers to clearer choices.
This is the modeling aspect. If you model the weather but never take into account barometric pressure, your model would fail if trying to determine the worst hurricane trajectories. These are the kinds of things you will be looking at in your decision models: searching for ways to explain the conditions that produce outcomes and to evaluate the strengths and weaknesses of the models you produce.
The three main ideas that your report should encompass are your ability to formulate a decision analysis research question based on an appropriate data set, develop your model, and finally evaluate the model's utilities, results, strengths, and weaknesses. In short, if your report fully encompasses these three concepts, you will produce an authentic document that would stand on its own in a professional setting.
Data sources to choose from: The included spreadsheet lists data sets used in previous sessions of DAT 520. Students found these data sets, prepared them for modeling on their own, and wrote excellent papers on the topics. Remember that your data set needs to be appropriate for modeling a discrete set of choices.
Either those choices are built into the model as categorical variables, or you will need to do some legwork by converting continuous variables into rational categorical groups. This activity would be part of the data preparation and documented in your data appraisal section.
Your final project must include the following sections:
Title Page
Abstract: 300 words or less
Table of Contents
Introduction, with research question: Up to two pages
Data Appraisal: Up to two pages
Techniques (a.k.a. Methods): Up to two pages
Evaluation: Two to four pages
Model, including optimizations
Up to one page for graphic(s)
Up to two pages for model explanation
Results: Up to two pages
Limitations: Up to two pages
Conclusion: Up to two pages
Sources: Note that the core elements add up to about 15-20 pages, double-spaced. The overall target for the core elements is still 15-20 pages, so that you have room to adjust each section according to the needs of the project. Everything you need to say in the report should fit within 15-20 double-spaced, 12-point font pages with one-inch margins.
To see some good final projects, consult the exemplars. Not all of them are 100% perfect papers, but they do embody the level of complex thinking that characterizes an interesting project.
The idea behind the page limit is to explore the concept of "less is more." If you add up the text, graphics, sources, and supporting material from all the milestones, you end up with 15 to 20 pages. For the final, that means some compression needs to occur.
This means finding the most important information from what you have previously written and leaving room for the new parts that you need to write. Follow the list of required elements for the final to guide how to structure your research paper.
Specifically, the following critical elements must be met in your final submission:
I. Introduction: Analyze the purpose, type, intended populations, and uses of the analysis to establish an appropriate context for the data-mining and analysis plan.
II. Data Appraisal
A. Characterize the data set. For example, what is the purpose such data are generally used for?
B. Appraise the data within the context of the problem to be solved and industry standards. How will you use the data? For example, expound upon the limitations of the data set in the context of your needs.
C. Explain the utilities that you will be using and how the data supports that choice.
III. Select Appropriate Techniques
A. Determine and explain the appropriate steps for preparation of the data sets into a usable form: what steps were taken to make data
descriptions clear, how extreme or missing values were addressed, and how data quality was improved.
B. Determine the appropriate steps (including: risk assessment, probability calculations, and modeling techniques) for data manipulation and in-depth analysis to support organizational decision-making.
C. Models and checkpoints: How will you optimize the models, what will you test for, and how will you build in checks to determine a successful analysis?
D. Defend the ethicality and legality of the analytic selections made for use, interpretation, and manipulation of the data based on industry
standards for legal compliance, policies, and social responsibility. If there are no potential ethical and legal compl iance issues, explain how your prep and use of this data are both ethical and legal.
IV. Defend and Evaluate Choices
A. Why are these choices the best for the data and problem at hand? What research or industry standards are supportive of your choices of methods? Explain how the methods chosen will support organizational decision-making.
B. Determine the agility of these choices for decision support based on research and relevant examples: how can they be adapted to alternative needs or reapplied to future analysis?
C. Address ethical and legal issues that might arise from the use and interpretation of the data, based on industry standards, policies, and social responsibility. How can you ensure that your selected procedures, use of data, and results will be socially responsible and in line with your own ethical standards?
D. Implement your plan: Perform data preparation, mining and modeling procedures, and create your decision support solution.
V. Decision Tree Model (bottom-up, top-down): Include the detailed process and programming steps necessary to complete the analysis. Be sure to:
A. Defend the overall structure and purpose of the tree model in organizational decision support.
B. Develop process-documentation that addresses potential complications. This piece should resemble a recipe/outline that provides enough
information for addressing potential implementation issues.
C. Evaluate the results of your decision tree model. At minimum, attend to the following:
1. Are the results reasonable?
2. How accurate is your model?
3. Are there missing or extraneous elements that could have influenced your results?
4. What common errors are made during creation of the model you chose? How did you ensure that you did not make these errors?
VI. Articulation of Response/Final Report: Utilizes visualization options that effectively address the needs of the audience. Options may include annotated shell tables, visualizations, and a compositional structure.
To guide you in writing your final paper, follow the Final Project Rubric. The rubric is less about format and more about thought. Specifically, you should write sections that detail the limitations and justification for your analysis. You should also take the time to address any ethical or legal issues that connect with your results or decisions being analyzed.
You should annotate and caption your graphics. You could include a table that characteri zes the data set. You should address what your model does to assist decision makers. You should defend your choices of variables and groupings. Lastly, you should address the agility of your analysis and how it might be applied to future uses.
Milestones
Milestone One: Choose a Data Set and Formulate Decision Analysis Research Question
In Module Two, you will choose a data set from the curated list of sources (Final Project Topics and Sources.xls), or you may submit your proposal for a different data source than those listed. Then you will write a decision analysis research question, which should be two to three pages in length and framed as a discrete set of choices to be analyzed. This milestone is graded with the Milestone One Rubric.
Milestone Two: Develop Decision Analysis Model
In Module Five, you will draft your decision tree. This task presupposes a data set, a viable decision analysis research question, and the necessary data prep.
To complete this milestone, you may have to experiment with different modeling styles. The main objective is to draft your model, explain what you did, and explain why it is the best model for your research question. This milestone is graded with the Milestone Two Rubric.
Milestone Three: Revise and Evaluate Decision Analysis Model
In Module Seven, you will revise and evaluate your decision model based on the feedback you received from the instructor for the previous milestone.
Evaluation in this case could mean a few different things. If you are performing a bottom-up style recursive partitioning analysis, you should report on the error rate and variable selection. You might also consider alternative variable categorizations to improve your model. If you are performing a top-down decision tree modeling exercise, what are the threshold values that cause the tree to flip?
You should perform sensitivity analysis on the critical variables in your tree and report what those sensitivity analyses are telling you. For either style of modeling, what makes your tree stronger? What bre aks the model? This milestone is graded with the Milestone Three Rubric.
Final Submission: Decision Analysis Model and Report
In Module Nine, you will submit your decision analysis model and report, compiling all the components used to develop the model and produce the report, as well as a leading abstract, table of contents, and in a format that addresses all of the critical elements in the instructions. The project should include sections that detail the limitations and justification for your analysis.
You will probably be compressing what you wrote for your introduction to make it fit within the eight-page limit. You should also take the time to address any ethical or legal issues that connect with your results or decisions being analyzed. Lastly, you should address the agility of your analysis and how it might be applied to future uses. This assignment is graded with the Final Project Rubric.
Guidelines for Submission: The final report will be a 15-20 page research paper, double-spaced, in 12-point Times New Roman font with one-inch margins all around and APA citations.
Title page, abstract, appendices and bibliography of sources are extra beyond the 15-20 pages of the report. You may include one page or less of annotated/captioned graphics as part of the report. The purpose of the limits is to keep the discussions compact and to maintain the integrity of publication-quality research.