Abstract Data modeling requires a sufficient sample size for reproducibility. A small sample size can inhibit model evaluation. A synthetic data generation technique addressing this small sample size problem is evaluated: from the space of arbitrarily distributed samples. a subgroup (class) has a latent multivariate normal characteristic; synthetic data can be generated from this clas... https://www.bekindtopets.com/save-big-Winslow-Baskets-2-sizes-best-find/