https://github.com/pdwaggoner/gasp2020

Repo for GASP 2020 talk
https://github.com/pdwaggoner/gasp2020

Last synced: 24 days ago
JSON representation

Repo for GASP 2020 talk

• Host: GitHub
• URL: https://github.com/pdwaggoner/gasp2020
• Owner: pdwaggoner
• Created: 2020-11-03T19:46:18.000Z (about 4 years ago)
• Default Branch: main
• Last Pushed: 2021-02-11T23:10:19.000Z (over 3 years ago)
• Last Synced: 2023-10-20T22:17:50.206Z (about 1 year ago)
• Language: R
• Homepage:
• Size: 1.07 MB
• Stars: 3
• Watchers: 2
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

        
# GASP 2020

Repo with data & code to accompany my GASP 2020 talk

Make sure you have the latest version of JDK installed to access the H2O engine. Download [here](https://www.oracle.com/java/technologies/javase/javase-jdk8-downloads.html). 

# Talk Abstract

Neural networks are increasingly popular and powerful machine learning models with remarkable predictive accuracy. Though some packages exist in R to build them (e.g., nnet), in general these types of one-off packages are limited, mostly due to the complexity of neural network topologies. As such, external machine learning engines exist to build, run, and test these types of advanced computational architectures at scale. For example, Keras and Tensorflow are powerful machine learning engines most often used in Python. But an API exists to allow for application in R using common R syntax. In this talk, I will walk the audience through the basics of neural networks, applications in the social sciences, and then the mechanics of constructing and visualizing neural networks using two widely used and powerful external machine learning engines: Keras//Tensorflow and H2O. As a part of the talk, I will interactively share hundreds of lines of code, which can be updated and repurposed for a host of applications after the talk. While every effort will be made to ease the audience into the complexities of neural network workflows using these engines, audience members who are comfortable in R and RStudio, as well as comfortable with machine learning terminology will likely get more out of the talk. Yet, all are welcome regardless of prior training, expertise, and interests.