Stan is an open-source statistical modeling language. The language was designed for the flexible specification of statistical models. Combined with advances in statistical computing has made it a popular choice, especially for Bayesian inference.

I'll discuss the motivation for Stan: hierarchical models. These models make sense: people are grouped in states, schools are in districts, etc. In many cases, traditional inference algorithms fail to estimate parameters correctly for these models.

If you work on modeling data, you will eventually run into limitations of whatever purpose-built package you're using. At that point your choices are to either use a flexible probablistic programming language or to write your own package.

I'll wrap up with a description of the type of models I work on for clinical trial data, with as much detail as I can fit in the time remaining.