Active models of travel, particularly walking, are an integral part of the multi-modal transportation system in urban areas. Walking provides numerous benefits at the individual and community levels (e.g., health benefits, reducing traffic congestion, emissions, and energy consumption). Nevertheless, safety concerns represent a major roadblock to the optimal utilization of walking as a key mode of travel. Pedestrians are among the most Vulnerable Road Users (VRUs) who are at a higher risk of being killed or severely injured as a result of road collisions. Previous research shows that many pedestrian behaviours could increase the risk of collisions significantly. Pedestrian violations, either temporal or spatial, stand as one of the riskiest behaviours that impact pedestrian safety. However, investigating such behaviour and quantifying its impact on safety are scarce in the literature. Accordingly, this research aims at developing a comprehensive framework to analyze pedestrian violations and understand when and where they can lead to collisions. To address these goals, the research utilized historical records of collisions that involve pedestrian violations. State-of-the-art statistical models (Copula models, Bayesian Structural Equation Modelling), Machine Learning techniques (Latent Class Analysis clustering), and Deep Learning methods (Self-Organizing Map) were applied to understand the factors contributing to such collisions on the micro-level (intersection and mid-blocks) and macro-levels (traffic analysis zones) and understand the characteristics of locations that experience a high frequency of those collisions. Additionally, a novel approach (dynamic R-vine copula-based time series model) was proposed to analyze the efficiency of pedestrian safety treatments that are implemented as part of vision zero programs. This approach enables the accurate assessment of the treatments, identifying the most effective combination of treatments, and investigating the association between area characteristics and treatment combination performance. Overall, this dissertation provides a solid understanding of pedestrian violations and safety for decision-makers, safety practitioners, and academia. / Thesis / Doctor of Philosophy (PhD)