Embodied Migration Analytics investigates how mixed-reality and embodied interaction can support collaborative exploration and interpretation of complex data on urban superdiversity, neighbourhood change, and migration in Canadian cities. Using an iterative human-centred design approach, the project develops and evaluates a multi-user immersive environment that enables researchers and stakeholders to navigate across spatial scales, compare places and time periods, combine multidimensional data layers, and collaboratively investigate patterns in urban diversity. The project uses Vancouver as its initial case and draws on the Superdiversity in Canadian Cities research to examine how interaction and representation techniques can make complex migration data more accessible, interpretable, and discussable without oversimplifying diverse communities. We also explore whether a grounded, locally deployable open-weight LLM can add value as a secondary conversational interface for navigating and interpreting the underlying data.