Extracting information from data, often also called data analysis, is an important scientific task. Statistical approaches, which use methods from probability theory and numerical analysis, are well-founded but difficult to implement: the development of a statistical data analysis program for any given application is time-consuming and requires knowledge and experience in several areas. In this paper, we describe AutoBayes, a high-level generator system for data analysis programs from statistical models. A statistical model specifies the properties for each problem variable (i.e., observation or parameter) and its dependencies in the form of a probability distribution. It is thus a fully declarative problem description, similar in spirit to a set of differential equations. From this model, AutoBayes generates optimized and fully commented C/C++ code which can be linked dynamically into the Matlab and Octave environments. Code is generated by schema-guided deductive synthesis....

Introduction Statistical approaches to data analysis, which use methods from probability theory and numerical analysis, are well-founded but difficult to implement: the development of a statistical data analysis program for any given application is time-consuming and requires knowledge and experience in several areas. AutoBayes [BFP99, FSP00] is a fully automatic high-level generator system for data analysis programs from statistical models which aims to overcome these barriers. AutoBayes follows the schema-based deductive approach to program synthesis. This means that the programs are constructed by the instantiation of generic algorithm schemas, e.g., EM. This process is supported by logic-based deduction to ensure the consistency between specification (i.e., statistical model) and synthesized program. AutoBayes uses a textual notation which is based on graphical models (more precisely, Bayesian networks) to specify the statist