Advanced Travel Demand Forecasting
We are developing a next generation activity-based microscopic simulation model for travel-demand forecasting across Metro Vancouver and the Okanagan region. The model is capable of simulation activities in the virtual and physical spaces, and their interactions. The workflow begins with a population-synthesis module that constructs a representative synthetic population consistent with regional

demographic and socio-economic controls. Empirical grounding comes from the British Columbia Activity Time Use Survey (BC-ATUS. These data capture the full spectrum of in-home, out-of-home, and virtual pursuits, ensuring the model reflects contemporary post-pandemic behaviours. Building on this foundation, a learning-enabled activity-generation and sequencing module employs deep and reinforcement learning to create coherent 24-hour schedules for every synthetic individual. The engine seamlessly integrates work, education, shopping, social, recreational, and at-home digital engagements, preserving behavioural realism while accommodating emerging telecommuting patterns. The resulting activity chains feed into destination-choice and mode-choice components calibrated to local land use, accessibility, and vehicle-ownership characteristics. A dynamic network-assignment layer then routes individual trips through time-dependent transport networks, updating travel times and iterating until demand–supply equilibrium is reached. This integrated, behaviourally rich architecture yields high-resolution forecasts that support evidence-based planning, climate-sensitive policy evaluation, and equity assessments. By uniting cutting-edge machine learning with detailed regional data, CliMR’s framework offers a next-generation tool for understanding and shaping the mobility future of British Columbia.