R Introduction to Machine Learning with tidymodels: Parts 1-2

This workshop is a 2-part series that runs from 9am-12pm each day:

• Tuesday, March 1
• Thursday, March 3

Machine learning often evokes images of Skynet, self-driving cars, and computerized homes. However, these ideas are less science fiction as they are tangible phenomena that are predicated on description, classification, prediction, and pattern recognition in data.

During this two part workshop, we will discuss basic features of supervised machine learning algorithms including k-nearest neighbor, linear regression, decision tree, random forest, boosting, and ensembling using the tidymodels framework.

To social scientists, such methods might be critical for investigating evolutionary relationships, global health patterns, voter turnout in local elections, or individual psychological diagnoses.

• Background on machine learning

  • Classification vs regression

  • Performance metrics

• Data preprocessing

  • Missing data

  • Train/test splits

• Algorithm walkthroughs

  • Lasso

  • Decision trees

  • Random forests

  • Gradient boosted machines

  • SuperLearner ensembling

  • Principal component analysis

  • Hierarchical agglomerative clustering

• Challenge questions

Prerequisites: D-Lab’s R Fundamentals or equivalent knowledge; previous experience with base R is assumed and basic familiarity with the tidyverse.

Workshop Materials: https://github.com/dlab-berkeley/Machine-Learning-with-tidymodels(link is external)

Software Requirements:Installation Instructions(link is external) for getting started with this working using R and RStudio.

Please register on the UCSF Registration for D-Lab Workshop portal here.