| Course | DBA-8060 Empirical Entrepreneurship Data, Technology, and Business Intelligence |
|---|---|
| Module | Module 4 |
| Paper type | Module 4 written assignment: data mining process, methods and algorithms |
| Length | About 518 words, 4 pages |
| Format | APA 7 student paper |
| School | Thomas Edison State University |
| Program | Doctor of Business Administration |
| Updated | October 2026 |
Free sample paper for DBA 8060 Module 4
Which Items Will Run Out in Three Weeks? A Data Mining Approach to Stockout Risk at DAKOKOO
Student Name
Doctor of Business Administration, Thomas Edison State University
DBA-8060: Empirical Entrepreneurship: Data, Technology, and Business Intelligence
Mentor Name
Month Day, Year
Which Items Will Run Out in Three Weeks? A Data Mining Approach to Stockout Risk at DAKOKOO
Introduction
DAKOKOO's buyers learn about stockouts when a retailer complains. A model that flags items likely to run out within three weeks would give them time to reorder or move stock between distribution centers. This paper follows a standard data mining process to build that model.
The Process
Fayyad et al. (1996) describe knowledge discovery as a sequence of steps: understanding the application and goal, selecting and preparing data, mining it and interpreting and acting on the patterns found, with frequent loops back to earlier steps. Sharda et al. (2018) present the closely related CRISP-DM cycle, which adds explicit stages for business understanding and deployment.
Business understanding came first. A useful prediction must arrive at least three weeks before the shelf empties, the time needed for an expedited reorder. The data came from 18 months of item-week records at three distribution centers, about 410,000 rows, with a flag for whether the item stocked out within the following three weeks. Predictors included on-hand weeks of supply, the supplier's lead-time variability, open purchase orders, recent sales growth and promotion flags from retailers. Preparation took most of the effort: 22 percent of rows lacked a receipt date and were completed from carrier records where possible.
Comparing Algorithms
I compared two classification algorithms. A decision tree splits the data on the predictor values that best separate stockouts from non-stockouts and produces rules a buyer can read. A random forest, developed by Breiman (2001), averages many trees grown on random samples of rows and predictors, which usually improves accuracy and reduces overfitting but makes individual predictions harder to explain.
Only 4 percent of item-weeks stocked out, so accuracy misleads: a model that predicts no stockouts at all would be 96 percent accurate and useless. I used the last six months as a test set and compared precision, the share of flagged items that did run out, and recall, the share of actual stockouts that were flagged.
| Model | Precision | Recall | Explainable to buyers |
|---|---|---|---|
| Decision tree (depth 5) | 0.41 | 0.62 | Yes, as rules |
| Random forest (300 trees) | 0.47 | 0.66 | Only through variable importance |
Choosing and Deploying a Model
The forest performed somewhat better, but the tree's rules, such as "weeks of supply below 2.5 and supplier variability above 12 days," let buyers understand and trust each flag. Because buyers must act on the list, I recommend deploying the tree first as a weekly watch list of about 150 items, with the forest run in parallel for six months. If the forest's added recall prevents enough stockouts to justify its opacity, it can replace the tree. Wang et al. (2016) note that analytics in supply chains often fails at the point of use rather than in the model, which is why deployment is planned with the buyers.
Conclusion
A disciplined process turned a vague complaint into a defined prediction task, a fair comparison and a model buyers can use. The model's value will be measured not by its test scores but by fewer stockouts in the next two quarters.
References
Breiman, L. (2001). Random forests. Machine Learning, 45(1), 5-32. https://doi.org/10.1023/A:1010933404324
Fayyad, U., Piatetsky-Shapiro, G., & Smyth, P. (1996). The KDD process for extracting useful knowledge from volumes of data. Communications of the ACM, 39(11), 27-34. https://doi.org/10.1145/240455.240464
Sharda, R., Delen, D., & Turban, E. (2018). Business intelligence, analytics, and data science: A managerial perspective (4th ed.). Pearson.
Wang, G., Gunasekaran, A., Ngai, E. W. T., & Papadopoulos, T. (2016). Big data analytics in logistics and supply chain management: Certain investigations for research and applications. International Journal of Production Economics, 176, 98-110. https://doi.org/10.1016/j.ijpe.2016.03.014
Reading the DBA 8060 Module 4 assignment instructions
This Module 4 page carries one of DBA-8060's five written assignments, which together count for 35 percent; the course site, not the public syllabus, places each assignment in a module. Module 4 covers the data mining process, methods and algorithms. Expect to frame a business question as a mining task, follow a recognized process such as CRISP-DM or knowledge discovery in databases, compare at least two algorithms and evaluate them with measures suited to the problem. Sharda, Delen and Turban's chapters on data mining give the process and algorithms. Choose a problem with a decision attached, so you can show how the model would be used, and explain your choice of evaluation measures when the outcome you predict is rare.
Inside the DBA 8060 Module 4 example
The paper begins with the buyers' problem: they hear about stockouts from customers. It sets the business requirement first, a warning three weeks ahead, and then walks through data selection and preparation with real volumes and the share of missing dates. Two algorithms are explained in plain terms, a decision tree and a random forest, and a test on the most recent six months compares them. A table reports precision, recall and explainability, and a paragraph shows why 96 percent accuracy would be meaningless at a 4 percent stockout rate. The paper deploys the readable tree first, runs the forest in parallel and judges success by fewer stockouts.
DBA 8060 Module 4 rubric: what earns full marks
Marks reward a mining process followed stage by stage, algorithms that suit the task, evaluation that fits a rare outcome, a real business decision behind the model and scholarly sources. Following each stage of a recognized process, including business understanding and deployment, shows method. Choosing evaluation measures for an imbalanced outcome, and explaining why, shows technical judgment. Weighing performance against explainability shows awareness of how models are used in organizations. Reporting results on a held-out test set, with figures, makes the comparison credible. Linking deployment to the people who act on the predictions shows the managerial perspective the textbook takes.
Common DBA 8060 Module 4 mistakes, and how to avoid them
Students lose points by listing algorithms without applying any, by reporting accuracy on a rare outcome or by ending with a model that no one will use. Some skip business understanding and start with the data. Others describe the steps of CRISP-DM without showing what happened in each. Start with the decision the model will support and the lead time it needs. Report data volume and preparation honestly. Compare two algorithms on a test set with precision and recall, or another measure fit for the problem, and explain your choice in terms of how people will act on the output. A free sample built around your own DBA-8060 data mining problem is available as a first draft. State the base rate of the outcome before reporting any model results.
Write yours, or have the desk draft it
This paper is an original model document written by our desk, not a submitted student paper and not an official Thomas Edison State University document. Read it for the moves, then write your own to the instructions in your classroom. If you want one built to your exact prompt and rubric, the first custom sample is free and arrives in 24 to 48 hours.
More DBA 8060 and Doctor of Business Administration sample papers
- DBA 8060 Module 1: Written Assignment: Big Data as an Entrepreneurial Opportunity at a Retail Distributor
- DBA 8060 Module 2: Written Assignment: Lead-Time Data, a Regression and Two Charts for DAKOKOO
- DBA 8060 Module 3: Discussion: A Star Schema and a Scorecard for DAKOKOO's Retail Contracts
- DBA 8060 Module 5: Written Assignment: Mining Supplier Emails and Shopper Posts for Lead-Time Signals
- DBA 8060 Module 6: Written Assignment: Optimizing Safety Stock and Simulating Expedite Costs
- DBA 8060 Module 7: Discussion: Batch, Stream or Managed Cloud? Choosing DAKOKOO's Big Data Tools
- DBA 8060 Module 8: Final Project: A White Paper on Implementing Big Data at DAKOKOO Distribution
- OML 8020 Module 4: Leaders and Culture Seminar Paper: How CEO Values and Behavior Shape Organizations
- DBA 8080 Module 4: Final Project: The Future of Assessing Adult Learners When AI Can Write the Answer
- DBA 8050 Module 3: Briefing Note: What Our Customer Survey Can and Cannot Tell Us
- DBA 8010 Module 7: Discussion: Labor Conditions on a Purple Carrot Farm
DBA 8060 Module 4 questions, answered
Where can I find a free DBA 8060 Module 4 sample paper?
The paper above builds a stockout prediction model for DAKOKOO Distribution following the data mining process.
What are the stages of CRISP-DM?
Business understanding, data understanding, data preparation, modeling, evaluation and deployment, repeated as needed.
Why is accuracy misleading for rare events?
A model that never predicts the rare outcome can score high accuracy while missing every case that matters.
What is the difference between a decision tree and a random forest?
A tree gives one set of readable rules, while a forest averages many trees for better prediction but less transparency.
What does precision mean in classification?
The share of cases the model flags as positive that really are positive.