Automated Statistical Model Discovery with Language Models

arXiv:2402.17879v1 Announce Type: new
Abstract: Statistical model discovery involves a challenging search over a vast space of models subject to domain-specific modeling constraints. Efficiently searching over this space requires human expertise in modeling and the problem domain. Motivated by the domain knowledge and programming capabilities of large language models (LMs), we introduce a method for language model driven automated statistical model discovery. We cast our automated procedure within the framework of Box's Loop: the LM iterates between proposing statistical models …

abstract arxiv automated capabilities constraints cs.cl cs.lg discovery domain domain knowledge expertise human knowledge language language models large language large language models lms modeling programming search searching space statistical type vast