Summer Research: Bayes and Pretty Maps
I'm a hardcore R junkie, fanboy if you must. I blame this obsession on being completely self taught in the R programming language. I've never had a formal class that uses R, no matter how much I tried to convince my teachers. I appreciate a community that is so strong thousands of data nerds like myself can slap around Google and find nifty tutorials, examples, and books to assist us on the quest to become real data scientists.
As tribute to future R users and a bow to those who came before me, I spent this summer hacking away at two tutorials. One on R graphing and mapping with ggplot2 and ggmap. The other on Bayesian Hierarchical Models with the packages MCMCpack and coda. I attempt to write in the style similar to the best tutorials I read. Slightly vague, but with examples and explanations that can be filled in by the help page for each package. I consider intermediate R users those who can read through a help page, but are still working on their own functions and packages. These tutorials are written for those intermediates. Both of these tutorials were written in knitr. If there is enough interest I certainly would not mind writing a tutorial on report generation in R.
Below you will find links and abstracts for each of the tutorials. Please contact me if you have any questions or comments.
The purpose of this paper is to be an introductory text for the R programming language’s graphing and mapping features. R is a free software programming language and software environment for statistical computing and graphics. While it performs similar functions to packages such as SAS and STATA, the systems have specific strengths and weaknesses. This paper uses ggplot2 and ggmap to show R’s graphing capabilities.
The purpose of this documentation is to present a practical understanding and implementation of a Bayesian Hierarchical model. Bayesian Hierarchical models allow analysts to account for endogeneity. A Bayesian Hierarchical model is a Bayesian network, a probabilistic graphical model that represents a set of random variables and their conditional dependencies via a directed acyclic graph . Bayesian Hierarchical models subset themselves by containing three or more levels of random variables or use latent variables. One level uses within-unit analysis and another level for across-unit analysis. Within-unit models describes individual respondents over time. The across-unit model is used to describe the diversity, or heterogeneity, of the units.