<p>Regional housing development decisions can have devastating effects if not properly curated to the needs of future demographics, whereby undesirable pathways are challenging to reverse without extensive retrofits or costly replacements. Subsequently, forecasting building stock future demands is key to guide effective urban planning and housing policy development. The latter needs to be driven by a <i>digital construction strategy</i>, whereby conventional, and often subjective, approaches are replaced by data-guided techniques, to realize <i>what</i> should be built (based on objective demand forecasting), as well as <i>how</i> such residences should be manufactured, assembled, built, assessed and, ultimately, managed. As the strategy’s first step, the influence of demographics on shaping building stock demands is acknowledged yet inadequately quantified in literature. To address this gap, this study develops a data-driven framework for forecasting building stock using demographic features. The utility and effectiveness of the framework in forecasting dwelling demands is shown using a case demonstration employing a comprehensive survey of Canadian household demographics. The developed two-stage classification pipeline was shown to enhance the model performance—highlighting the potential of further optimized classification models. The analysis results demonstrated that current housing construction plans are not aligned with demographics-guided housing needs. The adoption of the developed framework’s model is expected to empower planners and policymakers to align future construction decisions with expected demographic makeup, marking the first step of the proposed digital construction strategy. Planning decision-makers may also utilize the developed framework's model to forecast building stock demands based on different population and urbanization scenarios, in support of sustainable development through minimizing subjective planning decisions and wasted resources.</p> Graphical abstract <p></p>