As part of my master's course, this project is an attempt to combine new technologies such as machine learning, deep neural networks, data evaluation, and parametric architecture dataset creation. The generated synthetic data will be used as urban architecture forms for different analyses and simulations, which can then be evaluated. Thus, a machine learning model can be trained on well-performing spatial forms. The final result is based on user input of some parameters like the number of buildings and inhabitants and the desired square meters, the pre-trained model automatically proposes then one or more optimized urban configurations stored and displayed as IFC files.

TECHNIQUES : BLENDER | SVERCHOK | RADIANCE | RANDOM FOREST | PYTHON