Designing Possibilities: Predictions and Projections

prediction in city planning, Theodore Galanos, presents an architectural design method supported by wind prediction that aims at improving outdoor wind comfort on architectural and urban scales. The authors explored various design options in a case study in Kosice, Slovakia, integrating the wind as a factor into the form- fi nding process and predicting its effects. InFraRed , a machine learning wind prediction tool was coupled with computer fl uid dynamics (CFD) to validate the analysis and test its suitability. learning has enabled the ability to address subjective factors and make predictions using spatial semantic maps. The Rovenir and presents an investigation on cross-validation of deep generative methods of fl oor plan design and the output quality in relationship with the training process. The results presented indicate data-driven methods depend not only on the size of the sample and training instructions but also on the distribution of samples. The fi nal contribution is a guideline for the design and curation of a fl oor plan dataset.