Software

Robert B Shaffer (PhD Candidate, University of Texas at Austin) and I are pleased to introduce bigKRLS, which can easily identify heterogenous effects. Currently available on CRAN and Github, bigKRLS combines recent big data packages for size and C++ extensions for speed into a new Kernel Regularized Least Squares algorithm. With one command, start interacting with your bigKRLS results in your web browser with Shiny! At useR! 2016, we released a new version of the algorithm which estimates first differences with a much lower memory footprint; our companion piece which analyzes the 2016 US presidential election with bigKRLS has been accepted by Political Analysis.

kerasformula

Offers a high-level interface for Keras for R, allowing researchers to build neural nets in as little as a line of code while providing the majority of customizability of Keras in terms of model design and estimation; featured by Tensorflow for RStudio. I recently gave a short course on kerasformula; slides and demo code available here.

polyreg

Automate formation and evaluation of polynomial regression models. The motivation for this package is described in 'Polynomial Regression As an Alternative to Neural Nets' by Xi Cheng, Bohdan Khomtchouk, Norman Matloff, and Pete Mohanty (<arXiv:1806.06850>). Now on CRAN.

From Data to Document

My recent post outlining how to automate RMarkdown for similar data sets is featured on RViews and R-bloggers.