2024 Advance Auto Team

The Challenge

MSBA-BA graduate students Pratik Baderdinni, Gift Wanapongpaisarn, Graham Fike, Bobby Beal, and Jared Brossman worked to predict the performance of new store openings for Advance Auto. Given the reliance on precise data for successful store launches, they focused on creating a model that could accurately forecast store performance using company data. Their goal was to address the challenge of leveraging complex data sets to make effective and accurate predictions of the performance of hypothetical new stores.

The Solution

The team conducted a regression analysis to uncover meaningful patterns and correlations in the data, identifying key variables influencing store performance. Realizing that regression analysis alone did not provide optimal predictive accuracy, they explored various machine learning techniques and determined that a neural network model was the best option. This model used machine learning to forecast whether a new store would perform above or below the median gross margin levels of all stores, achieving a training accuracy of 77.5% and a testing accuracy of 73%.

To demonstrate the financial benefits of their model, the team performed an economic analysis using a discounted cash flow approach. This analysis showed a significant potential gain in revenue for Advance Auto over five years, highlighting the model’s value in financial planning. Their recommendations included using the model to select new store locations, tracking store performance against a baseline, and refining the model with real-time data to ensure ongoing accuracy. This practical and innovative approach provided Advance Auto with a clear method for evaluating and improving new store performance.