This research, conducted in collaboration with Dr. Yonatan Dortheimer from the School of Architecture at Ariel University, focuses on developing a machine learning-based predictive tool for analyzing pedestrian movement in urban environments. Understanding walking patterns is essential for designing sustainable, accessible, and inviting cities. However, in practice, urban planners often lack accessible, data-driven tools to help them interpret pedestrian dynamics and incorporate relevant insights into decision-making processes. The project integrates applied systems analysis—through a Random Forest model that predicts pedestrian flow based on urban spatial features—with real-world empirical experimentation. Data is primarily collected from open sources and analyzed using feature reduction techniques and SHAP (SHapley Additive exPlanations) to interpret the contribution of different variables. The model is further enhanced by integrating data from multiple cities to ensure broader generalizability. For the experimental phase, we are developing an interactive GIS-based prototype to be tested in an urban design studio involving architecture students. The goal is to examine how designers interact with predictive tools, how these tools influence planning decisions, and how their usability and effectiveness can be improved. This research demonstrates how combining machine learning, applied systems analysis, and simulated environments can provide an effective foundation for decision-making in complex urban systems.