Typically, every city boasts its unique backdrop, characteristics, and challenges; however, urban difficulties in India exhibit similarities to those encountered in developed nations. The urban landscape of the country is marked by highly diverse traffic patterns, often lacking in lane discipline. Vehicles not only interact with those directly ahead but also with nearby ones, necessitating a deep understanding of lane-changing dynamics and their impact on traffic flow. Analyzing driver behavior in such scenarios, especially along mid-block segments of urban roads, presents significant complexity. This research focuses on extracting vehicular trajectory data from mid-block locations under varying traffic conditions to identify patterns in vehicle following and lane-changing behavior. The primary goal is to develop a driver behavior model that considers safe speed differentials between leading and following vehicles, along with the necessary clear space to prevent rear-end collisions. Additionally, it seeks to establish guidelines for maintaining safe lateral distances and speeds to reduce side-swipe collisions. This entails determining Time to Collision (TTC) thresholds based on following vehicle speeds and Safe Lateral Distance (SLD) thresholds using k-means clustering aided by a Python tool. Subsequently, risk levels are categorized into high, medium, and low tiers. The modeling phase employs Gradient Boosting Regression (GRB), Support Vector Machines (SVM), and Random Forest Algorithm (RFA) techniques.