Data Train: Introduction to Machine Learning

This is the workshop component to the Machine Learning workshop in the Data Train 2025 series.

1 Workshop Details

• 4 * 3-hour blocks, each ~90min theory, short break, ~90min hands-on
• Hands-on exercises on local devices, bring a laptop!

1.1 Day 1

• Theory / practice: 9:00 - 12:00
  • k-Nearest-Neighbors
  • General concepts
  • Decision Trees
• Break: 12:00 - 13:00
• Theory / practice: 13:00 - 17:00
  • Random Forest
  • Model evaluation
  • Boosting

1.2 Day 2

• Theory / practice: 9:00 - 12:00
  • Support Vector Machines (SVM)
  • Hyperparameter Tuning
  • Artifical Neural Networks
• Break: 12:00 - 13:00
• Theory / practice: 13:00 - 17:00
  • Specific endpoints
  • Variable Importance
  • Discussion

2 Quick Start Instructions:

We will use R version 4.5.1, but recent versions >= 4.1 should still work.

1. Get the materials: In an R session, install the {usethis} package and enter

usethis::create_from_github(repo = "https://github.com/bips-hb/datatrain_workshop_ml.git")

This will create a new project with the workshop materials on your local machine, and it will also fork the repository to your GitHub account (you can suppress this with fork = FALSE but there’s no harm in forking).

1. Install dependencies: Open the project and run renv::restore(prompt = FALSE) to install required R packages.

2. Verify: If the following example code produces a plot, you’re probably good to go:

library(mlr3verse)

#> Loading required package: mlr3

rr <- resample(
  tsk("sonar"),
  lrn("classif.ranger", predict_type = "prob"),
  rsmp("cv", folds = 3)
)

#> INFO  [23:22:53.306] [mlr3] Applying learner 'classif.ranger' on task 'sonar' (iter 1/3)
#> INFO  [23:22:53.651] [mlr3] Applying learner 'classif.ranger' on task 'sonar' (iter 2/3)
#> INFO  [23:22:53.706] [mlr3] Applying learner 'classif.ranger' on task 'sonar' (iter 3/3)

autoplot(rr, type = "roc")

(You’ll learn what that piece of code does in the workshop :)

If renv gives you trouble, see below for manual R package installation instructions.

3 Setup Instructions

1. Install R for your platform: https://cran.r-project.org/

• Installation instructions depend in whether you’re using Windows, Linux (whichever flavor), or macOS.
• We assumethe most recent R version, but all recent versions should work fine.

2. If you have neither Positron nor RStudio installed, install wither (Positron probably preferred)
3. Create a local copy of this workshop repository (https://github.com/bips-hb/datatrain_workshop_ml.git), using any one of these options (use whichever you are most familiar with):

1. Using R and the usethis package: usethis::create_from_github(repo = "https://github.com/bips-hb/datatrain_workshop_ml.git")
2. Running this in the terminal: git clone https://github.com/bips-hb/datatrain_workshop_ml.git
3. Using RStudio’s New Project -> Version Control dialog to clone the repository (analogously for whichever editor you’re using).

1. Install R packages required for the workshop by opening the workshop repository in Positron or RStudio and run renv::restore(prompt = FALSE).
   {renv} will automatically install all R packages with the correct versions listed in renv.lock.

In some cases, installation with {renv} might fail, and if that happens move on to the next section to install packages manually.

3.1 Manual package installation instructions

Click to expand instructions

You should only need to install all packages manually if you were not able to use renv to install them automatically. (Or if you’re trying to get this code to run in a different environment than this repository)

If you want to “disable” renv so you can manually install packages, open .Rprofile and comment out the following line:

source("renv/activate.R")

The you can directly use {pak} for installation, which will try to automatically install system dependencies on Linux (see next note) if possible:

packages <- c(
  # Data
  "palmerpenguins", "mlr3data",
  # Learner backends
  "ranger", "xgboost", "kknn", "rpart", "e1071", "randomForest",
  "mlr3verse", "mlr3filters", # installs "mlr3", "mlr3learners", "mlr3viz", "mlr3tuning" ...
  "precrec", # ROC plots via mlr3, not auto-installed with mlr3viz
  # Viz / interpretability
  "rpart.plot", "effectplots",
  # Plotting / infrastructure, goodies
  "rmarkdown", "ggplot2", "patchwork", "usethis", "dplyr", "purrr", "ragg"
)

# Installing pak for faster package installation
install.packages("pak")

# Install packages if not available already
pak::pak(packages)

3.1.1 Linux Note

Click to expand

If you’re working on a Linux distribution such as Ubuntu (or something Ubuntu-based), you may have to install some system packages with sudo apt-get install ... beforehand.

For Ubuntu it would look like this, which you can run in the terminal of your choice:

sudo apt-get install -y git
sudo apt-get install -y libcurl4-openssl-dev
sudo apt-get install -y libfontconfig1-dev
sudo apt-get install -y libfreetype6-dev
sudo apt-get install -y libfribidi-dev
sudo apt-get install -y libgit2-dev
sudo apt-get install -y libglpk-dev
sudo apt-get install -y libgmp3-dev
sudo apt-get install -y libharfbuzz-dev
sudo apt-get install -y libicu-dev
sudo apt-get install -y libjpeg-dev
sudo apt-get install -y libpng-dev
sudo apt-get install -y libssl-dev
sudo apt-get install -y libtiff-dev
sudo apt-get install -y libxml2-dev
sudo apt-get install -y make
sudo apt-get install -y pandoc
sudo apt-get install -y zlib1g-dev

4 Further Reading

4.1 Code examples

We rely on the mlr3 framework and its free online book for the hands-on part of the workshop:

• Bischl, B., Sonabend, R., Kotthoff, L., & Lang, M. Applied Machine Learning Using mlr3 in R. (CRC Press, 2024)

4.2 Free Lectures (online with slides + videos)

Lecture materials take inspiration from these free and open-source lectures:

• Introduction to Machine Learning (“I2ML”): https://slds-lmu.github.io/i2ml
• Interpretable Machine Learning: https://slds-lmu.github.io/iml

4.3 Textbooks

• An Introduction to Statistical Learning: with Applications in R. (Springer, 2013).
• Hastie, T., Tibshirani, R. & Friedman, J. The Elements of Statistical Learning: Data Mining, Inference, and Prediction, Second Edition. (Springer, 2009).
• Murphy, K. P. Machine Learning: A Probabilistic Perspective. (MIT Press, 2012).
• Bishop, C. M. Pattern Recognition and Machine Learning. Information Science and Statistics. (Springer, 2006).
• Molnar, M. Interpretable Machine Learning: A Guide for Making Black Box Models Explainable (2023)