Chapter 2 Unsupervised analyses

As introduced in the previous section, the term unsupervised analysis refers to that critical part of the analytic phase where we only focus on the exposures, trying to characterize, explain, and describe the complex environmental mixture of interest. This could even be the ultimate goal of the analysis (as a matter of fact, to respond to common questions such as “what are the most common exposures in our populations?” or “can we identify subgroups of exposures that are commonly found together?” we do not to account for the outcome. In other settings, this will still be an important preliminary phase that will inform subsequent analytic steps.

Note that the focus here is not on understanding biological mechanisms through which chemicals or pollutants operate in the body. The focus of unsupervised analysis in this context is instead a purely epidemiologic one. By attempting to identify clusters of exposures without accounting the their relationship with a given outcome we are in fact grouping exposures based on features such as population distribution and shared sources rather than on similar mechanisms of action.