Modeling of uncertain geometry of cold formed steel members based on laser measurements and machine learning

•A framework of machine learning is invented for uncertain modeling of as-true CFS members based on laser measurements.•An optimization-based feature recognition algorithm is developed to characterize geometric features from laser-based point cloud models.•The multi-dimensional Gaussian process is applied to model the stochasticity and correlations of geometric features.•The as-true geometries of cold-formed steel is generated from the statistic machine-learning model by sampling of geometric features.•Finite element analyses are conducted and compared with different geometric imperfections modeling methods.