Level of Education of Students Involved
Graduate
Faculty Sponsor
Haydar Cukurtupe
College
Other
Discipline(s)
Computer Science, Information Technologies
ORCID Identifier(s)
0000-0002-4670-4877
Presentation Type
Poster Presentation
Symposium Date
Spring 4-24-2025
Abstract
Retail businesses rely heavily on data-driven decision-making, yet extracting valuable insights from raw sales data can be quite challenging. In collaboration with Family Express, we created an AI-driven analytics platform that utilizes data science algorithms to convert sales data into valuable business intelligence.
We began the analysis with data preprocessing and cleaning, where missing values were handled using forward filling, backward filling, and interpolation methods. To further reduce any inconsistencies in the data, techniques like window smoothing were used. A correlation analysis was performed to examine relationships between gas sales and in-store purchases across all locations, providing insights on how fuel transactions effect inside-store sales.
Time series analysis was conducted on coffee sales, with forecasting implemented using ARIMA and Exponential Smoothing models to predict future demand for all stores. This analysis helps Family Express plan their inventory. Additionally, a store performance comparison was carried out, evaluating impact of CTO activations (opening of hot food items inside store) on sales across locations to identify high-performing stores and areas requiring operational improvements.
Large Language Models (LLMs) were run locally to not send any sensitive data to third-party organizations like OpenAI and to generate AI-ready prompts, enabling smart data interactions and simplifying complex business queries. This allows end-users without technical expertise to get insights through natural language interactions.
Finally, by integrating correlation models, time series forecasting, and AI-driven insights, this platform equips Family Express with data backed strategies to enhance operational efficiency, improve sales forecasting accuracy, and support informed decision-making.
Recommended Citation
Varri, Bhavaj Madev; Cukurtepe, Haydar; Chamakuri, Manisai; Marri, Prathyusha; Gutti, Srikrupa; and Pasupuleti, Chinmai, "AI-Driven Sales Data Analysis for Family Express" (2025). Symposium on Undergraduate Research and Creative Expression (SOURCE). 1463.
https://scholar.valpo.edu/cus/1463
Biographical Information about Author(s)
I Bhavaj Madev Varri, along with Manisai Chamakuri, Srikrupa Gutti, Chinmai Pasupuleti, and Pratyusha Mari, are graduate students in Information Technology at Valparaiso University, set to graduate in Spring 2025. Under the guidance of Professor Haydar Çukurtepe, we are part of the data science team for this project. Our backgrounds in machine learning, analytics, and business intelligence led us to explore AI-driven sales data analysis. Through this project, we have gained hands-on experience in data preprocessing, time series forecasting, and correlation modeling to extract meaningful insights for Family Express. We are passionate about leveraging data science to drive informed decision-making, and this project aligns with our future goals of building intelligent solutions that enhance business efficiency and innovation.