I presented yesterday on Data Science-Ish: The Use of Mixed Effects Models for Analyzing Complex Data at the KnoxData group.
Here is an abstract for the talk:
Mixed effects (or multilevel) models are extensions of linear models (regressions) that are commonly used in experimental research and policy and evaluation contexts, but are less widely-used by data scientists. In this talk, I aim to introduce mixed effects model as a data science-ish statistical method that can be useful and is easy to estimate and interpret. I focus on two examples using the lme4 package in R: one each that is focused on traditional research (modeling hierarchically nested data) and commercial (estimating the best product when products have different numbers of reviews) uses. Finally, I discuss the close connection between mixed effects models and Bayesian methods. This talk should be of interest to data scientists and academic researchers (and those somewhere in between or in related fields!).
Slides are available here: https://jrosen48.github.io/data-science-ish
Video of the talk is available here: https://www.youtube.com/watch?v=1YY2FoCFIm4
Code and Libraries Used
The code used in the presentation is available here: https://github.com/jrosen48/data-science-ish