The project provides an opportunity for businesses and learners to collaborate to identify and translate a real business problem into an analytics problem. 

The projects, which can be short, will allow the student to apply predictive modeling techniques as well as related statistical and visualization tools for data mining to address the sponsors business problem.  The projects should cover common machine learning techniques that are focused on predictive outcomes and evaluate the performance of the prediction models and how to improve them through time. Some examples are:

• Application of key machine learning (ML) terminology, ML applications and distinguish from more basic analytics and big data techniques
• Implement machine learning functions
• Formulate and communicate (orally, written) advanced analytics concepts
• Demonstrate ethical and professional standards related to the field of data analytics

You should submit a high-level proposal/business problem statement including relevant data sets and definitions, a list of acceptable tools (if applicable), and expected deliverables. Business datasets could be provided based on a non-disclosure agreement or in an anonymized/synthetic data format that is relevant to your organization and business problem. The  course instructors will review the documents to confirm the scope and timing of the proposed problem and its alignment with the capstone course requirements.

Analytics solution may be applicable for (however they are not limited to) the following topics:

1. Demand for social services (healthcare, emergency services, infrastructure, etc.)
2. Customer acquisition and retention
3. Merchandising for trade areas (categories)
4. Quantifying Customer Lifetime Value
5. Determining media consumption (mass vs digital)
6. Cross-sell and upsell opportunities
7. Develop high propensity target markets
8. Customer segmentation (behavioral or transactional)
9. New Product/Product line development
10. Market Basket Analysis to understand which items are often purchased together
11. Ranking markets by potential revenue
12. Consumer personification

To ensure students’ learning objectives are achieved, we recommend that the datasets are at least 20,000+ rows in size. Data need to be ‘clean’. If more than one database is provided, which must be conjoined, students will be required to integrate them. This supports the learning experience and minimizes partner data preparation.