Kenny ShirleyStatistician
I’m a research scientist in the demand forecasting group at Amazon in New York City. My research interests (past and present) include hierarchical Bayesian modeling, MCMC methods, data and model visualization, text mining, and other topics related to applied statistics.
This is my personal site, which is a mix of statistics research, side projects (mostly sports-related) and other stuff.
SOFTWARE
- R package summarytrees:
This is an R package (co-written with Howard Karloff) to compute and visualize maximum entropy summary trees. If you have a large, node-weighted tree (1000's of nodes, or more), our package aggregates the nodes in an optimal way and interactively visualizes the summary tree (typically 100s of nodes) using d3.js.- [1] Here is a link to the GitHub repo.
-
[2] Here are two vignettes describing a data analysis using the package:
- Analyzing the Gauss subtree of the Mathematics Genalogy Project.
- Analyzing the hierarchy of the webpages listed in the DMOZ directory of webpages
- [3] Here are a few examples of the interactive visualization:
- [4] Here is the original paper describing the work (from EuroVis 2013).
- R package LDAvis:
The LDAvis package (co-written with Carson Sievert) provides an interactive visualization of the topics learned from LDA. It is designed to allow users to easily interpret topics, by showing, for each topic, a ranked list of the most relevant terms, where relevance is a combination of a term's frequency within and exclusivity to a topic.- [1a] Here is a link to the Github Repo.
- [2] LDAvis on CRAN.
- [3] Here is an example of using LDAvis to visualize the topics learned from a set of movie reviews.
- [4] Here is a demo of LDAvis on some old news articles.
- [5] Here is the original paper describing the work (ACL workshop ILLVI 2014).