Learning to Predict: The Case of Smartwatch Adoption

Overview: 

This case is used to teach students to use Machine Learning (ML) and Predictive Analysis (PA) to predict the smartwatch adoption by consumers. It covers the classification machine learning algorithms (1) Lasso, (2) Ridge, and (2) Elastic Net Regression. This case is performed on a dataset that is loaded and run in R, a popular programming language used for statistical analysis and machine learning. 

Learning Objectives:

• Use the programming language R

• Supervised vs. unsupervised machine learning (ML)

• Train vs. test data

• Bias-variance tradeoff

• Overfitting

• Model selection

• Understand and use to following classification algorithms:

1. Lasso

2.  Ridge

3. Elastic Net Regression

Usage:

Here is one suggested way to teach the case and its concepts:  

•   Discuss the case and the case concepts in class. 

•   Students then complete the case materials. 

•   Finally, the solution file is discussed in a subsequent class.

Target Classes:

Graduate students in any business discipline

Experience Level:

Advanced Beginners or Competent

Files Titles and Explanations: 

1. "Instructional Materials.pptx": Use this slide deck to present the key concepts (i.e., Lasso, Ridge, Elastic Net) to students in class. Plan for about 30-40 minutes of lecture introducing the key concepts.

2. "S_Case Document.docx": Use this file in the class to discuss the problem and explain to students what needs to be done in the case. This file can also be posted on learning management systems (LMS), such as Canvas as the case. It includes all the tasks and exercises that need to be completed by the student. 

3. "Case Solution.Rmd": This file is instructor-only and includes the solution code along with the outputs and results of the analysis. It is an R Markdown file that instructors can run the code chunk by chunk for demonstrating solutions in class.

4. "Case Solution.html": This file is instructor-only and is the knitted html version of the Case Solution.Rmd file in case the instructor wants to show the solution without running it in R. It can be used as a supplement to the .Rmd file. Instructor can also demonstrate how .Rmd files can be exported to .html for ease of use by non-R users or readers.

5. "Instructor Orientation.docx": This file is an extended version of the case question file "The Case of Smartwatch Adoption_Students.docx". It also contains additional resources for the instructors and guide to related dataset, data dictionary, etc.

6. "S_Data.csv" - This file contains the dataset for the case, which will be loaded in R. 

7. "S_Data Description Sheet.txt" – This file contains detailed information about the variables used in this case.

Installations: To work on this case, install R and R Studio from the CRAN project website.

Contact: For any questions about the case or the files, please email unnati@illinois.edu.