Modeling-Visitor-Travel-Patterns-to-Restaurants-in-Urban-Environments

🚀 Project Overview
This research project is an intelligent modeling framework developed to understand and predict visitor traffic patterns to restaurants in urban areas. The project was developed as part of a graduate research initiative at San Diego State University and combines advanced deep learning and econometric modeling to enable data-driven decisions for smart city planning, restaurant operations, and urban mobility systems.

📄 Read the full project report:
📘 Project Report (Google Docs)

💻 Access the project code:
💾 Source Code (Google Drive)

---

📊 Problem Statement

Urban foot traffic to restaurants is shaped by a range of factors including restaurant quality, popularity, distance, and dwell time. This project aimed to:

• Forecast hourly restaurant visitor demand
• Classify restaurants into popularity tiers
• Simulate visitor decision-making using interpretable behavior models

---

🧠 Methodology

🔹 LSTM (Long Short-Term Memory)

• Used for time series forecasting of hourly restaurant demand
• Achieved 87.82% accuracy
• Enables use cases like staffing optimization and delivery planning

🔹 Kolmogorov–Arnold Networks (KAN)

• Used for classifying restaurants into Low, Medium, and High popularity tiers
• Achieved 94.72% classification accuracy
• Supports marketing segmentation and performance analysis

🔹 Discrete Choice Modeling (DCM)

• Modeled visitor choice behavior using a multinomial logit framework
• Identified key decision factors:
  • Quality (+0.57), Dwell Time (+0.39), Distance (–1.20), Popularity (–0.31)
• Offers actionable insights for location strategy and consumer behavior analysis

---

🛠️ Tools & Technologies

• Languages: Python
• Libraries: TensorFlow, PyTorch, PyKAN, Statsmodels, Scikit-learn, Pandas, NumPy, Matplotlib
• Techniques: LSTM, KAN, Multinomial Logit Model (DCM), SMOTE for class balancing, Feature Engineering, MinMaxScaler and StandardScaler for normalization

---

🌍 Real-World Applications

• Restaurant Operations: Forecast rush hours, optimize resource allocation
• Urban Planning: Inform zoning and public transit adjustments
• Mobility Services: Enhance delivery routing and surge pricing
• Retail Analytics: Understand traffic patterns to support site selection and strategy

---

📈 Key Highlights

• Built an interpretable AI pipeline combining deep learning and behavior modeling
• Processed and modeled 6K+ mobile location records for analysis
• Delivered insights on visitor behavior and demand patterns to guide business and city infrastructure decisions