Welcome to a small little app that is meant to provide a quick example of how a simple Bayesian machine learning model can be implemented in the area of voter analysis. Specifically this app leverages a collaborative filtering method rooted in matrix factorization, as helpfully abstracted by the Likely.js library for Node https://www.npmjs.com/package/likely. So, while this isn't a perfect app, I hope that the fundamental application of machine learning concepts to something as important as the election of the U.S. President is made more clear.
- Fork and clone this repo
- Open db/seedData.js and adjust the NUM_VOTERS constant to your desired number of voters!
npm install
npm start
- Go Predict Voter Issue Sentiment!
This is an obviously over-simplified version of how machine learning has made its way into the predictive analytics used by political campaigns. For a brief and introductory overview of this topic and to see this app in action, please watch my Fullstack Academy tech talk here: https://www.youtube.com/embed/SZY6_3nS4WM
I'd love to hear your feedback. I'm best reached via mw [at] mikewill.net.