Chick-fil-A Corporation Collaboration (Spring 2024, Fall 2024)
In the project with Chick-fil-A, the students compared multiple methods for creating synthetic datasets utilizing four different commercially available datasets. In general, organizations like Chick-fil-A and others increasingly rely on large volumes of data to inform strategic decision-making using methods such as predictive analytics or time series forecasting. However, for many companies, this type of modeling might be outsourced to partner companies specializing in this type of analysis. Partnering with vendors raises additional concerns related to data privacy, security, and regulatory compliance which have posed significant challenges in sharing and utilizing real-world data. Data breaches, unauthorized data access, and the potential misuse of sensitive information have heightened the urgency for solutions that safeguard privacy while enabling data utility.
The analysis the students performed was two-fold. First of all, the students compared the descriptive characteristics of the generated synthetic data to the same characteristics of the original data. They asked questions such as if the distributions and correlations were similar, etc. For synthetic data generation methods which seemed promising based on the descriptive characteristics of the data, the students then compared the performance of predictive models. They compared the results one would infer from the models when using the synthetic data as compared to the results inferred from the original data.
Students worked across two semesters on this project in teams. The resulting final presentations from each semester can be found below. Furthermore, two master's theses resulted from this initial project collaboration which can be found at these links:
Kofi Sarfo, “Investigating the Privacy-Utility Trade-off in Synthetic Data Generation using Correlated Attribute Mode”: https://dc.etsu.edu/etd/4537/
MD Raisul Islam Roni, “Postprocessing GAN-Generated Synthetic Time Series Using Dynamic Time Warping”: https://dc.etsu.edu/etd/4635/
