Wrap-up: Where to go from here

Lecture 23

Dr. Benjamin Soltoff

Cornell University
INFO 4940/5940 - Fall 2024

December 5, 2024

End-of-semester logistics

Remaining assignments

• Homework 05
• Homework 06/extra credit
• Group project

Build a simple machine learning stack

RStudio Workbench

• Access to RStudio Workbench and Google Cloud Storage will end at some point after December 20th
• All INFO 4940/5940 materials remain available in your repos on GitHub as long as you are an active student
• Any other work you have done on the server will not be accessible after the end of the semester
• Where will you go from here?

Software installation

Programming language

• R
• Python

IDE

• RStudio IDE
• VS Code
• Positron

Reproducibility

• Quarto
• Git
• Docker

To renv or not to renv

Benefits

• Isolated
• Portable
• Reproducible

Drawbacks

• It’s a pain to configure for every project
• Some packages have issues installing via {renv}

Install some core R packages

# install the major packages from the course published on CRAN
install.packages(c(
  "tidyverse", "tidymodels", "devtools", "usethis",
  "styler", "keras3", "ranger", "bonsai"
))

# install a package hosted on GitHub
remotes::install_github(repo = "cis-ds/rcis")

Create a GitHub account

• Ordinary GitHub
• GitHub Enterprise for Cornell

Configure Git

usethis::use_git_config(
  user.name = "Your name", 
  user.email = "Email associated with your GitHub account"
  )

Painless authentication with PAT

Personal Access Token

• Uses HTTPS protocol
• Integrates with the usethis package to automate some Git workflows
• Alternative: use SSH

Happy Git and GitHub for the useR

Setup PAT authentication

• GitHub.com
• GitHub Enterprise

Create PAT

usethis::create_github_token(
  scopes = c("repo", "user", "gist", "workflow"),
  description = "<DESCRIBE YOUR DEVICE>"
)

Store PAT

gitcreds::gitcreds_set()

#> ? Enter password or token: ghp_xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
#> -> Adding new credentials...
#> -> Removing credentials from cache...
#> -> Done.

Create PAT

usethis::create_github_token(
  scopes = c("repo", "user", "gist", "workflow"),
  description = "<DESCRIBE YOUR DEVICE>",
  host = "https://github.coecis.cornell.edu/"
)

Store PAT

gitcreds::gitcreds_set(url = "https://github.coecis.cornell.edu/")

#> ? Enter password or token: ghp_xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
#> -> Adding new credentials...
#> -> Removing credentials from cache...
#> -> Done.

What have you learned?

Learning objectives for INFO 4940/5940

• Train and evaluate machine learning models using a variety of algorithms.
• Collect and wrangle data for machine learning.
• Deploy machine learning models in a production environment.
• Communicate results of machine learning analyses to a non-technical audience.
• Implement reproducible machine learning workflows using version control and literate programming.

Where to go from here

Courses

Foundational (much more theory and math)

• CS 3780: Machine Learning for Intelligent Systems
• ECE 3200: Fundamentals of Machine Learning
• ORIE 3741: Learning with Big Messy Data
• STSCI 3740: Data Mining and Machine Learning
• INFO 3950: Data Analytics for Information Science

Domain applications

• INFO 3350/6350: Text Mining History and Literature
• INFO 3370/5371: Studying Social Inequality Using Data Science
• INFO 4100/5101: Learning Analytics
• INFO 4300: Language and Information
• INFO 4940: Advanced NLP for Humanities Research
• INFO 4940/6940: How LLMs work, their potential and limitations

Find a community

Online communities

• Data Science Learning Community
• R Ladies
• Posit Community

Keep your skills fresh

• Integrate reproducible workflows into other courses/projects
• Tidy Tuesday
• Create a blog