hadley/rmdl

rmdl is an R package that helps researchers manage multiple statistical models at once, with a focus on causal and epidemiological analysis. Instead of writing and running one model at a time, it lets you define a set of variables and their roles, like which is the exposure, which is the outcome, and which are other factors, and then automatically generates and fits several models in parallel. The package works through three main ideas. First, it lets you assign roles to your variables, so you can clearly mark what each one represents in your analysis. Second, it generates formulas for you, you describe a pattern of relationships, and it creates the corresponding statistical models. Third, it organizes everything into a tidy table where each row represents one fitted model, complete with metadata about what was included. The example in the README shows this clearly: using the built-in mtcars dataset, a single formula specification produces six models at once, covering two outcomes (mpg and hp) against the same exposure (weight) with other variables mixed in. The resulting table summarizes each model's formula, outcome, exposure, and fit status in one organized view. This would appeal to epidemiologists, public health researchers, or analysts who regularly run many variations of regression models and need to track which variables played which roles across each one. The package is marked as experimental, so it's still in active development and may not be production-ready. The README points to vignettes for deeper documentation but doesn't provide extensive detail beyond the core example.