McGill Initiative in Computational Medicine

Introduction to Supervised Machine Learning (Tabular Data)

This repository contains the full materials for the MiCM Workshop: Introduction to Supervised Machine Learning, designed to teach fundamental supervised learning concepts through hands-on, medically themed exercises (simulated) using tabular data.

Workshop Overview

Audience: Researchers, trainees, students interested in supervised learning. Duration: ~2 hours. Focus: Regression, classification, regularization, nonlinearity, and MLPs for medical tabular data. Tools: Python, pandas, scikit-learn, and matplotlib.

The workshop consists of:

• Guided exercises implemented in Jupyter notebooks.
• Synthetic medical datasets that simulate realistic relationships.
• A final mini-hackathon applying all learned concepts.

Google Colab links

You don't have to install everything for today's workshop.

Notebook	Colab Link
00 Intro
01 Regression
02 Regularization
03 Nonlinearity
04 MLP
05 Hyperparameters
06 Hackathon

Repository Structure

QLS-MiCM_Introduction_to_supervised_machine_learning-main/
│
├── Outline/
│   └── workshop_outline.md
│
├── Slides/
│   ├── 00_intro.md
│   ├── 01_linear_regression.md
│   ├── 02_regularization.md
│   ├── 03_nonlinearity.md
│   ├── 04_mlp.md
│   ├── 05_training_eval.md
│   └── 06_hackathon.md
│
└── Exercises/
    ├── data/
    ├── scripts/
    │   └── generate_datasets.py
    ├── workshop_utils.py
    └── solutions/
        ├── 00_intro_and_data_exploration.ipynb
        ├── 01_linear_regression_medical.ipynb
        ├── 02_regularization_high_dim_medical.ipynb
        ├── 03_feature_engineering_and_nonlinearity.ipynb
        ├── 04_mlp_for_tabular_medical.ipynb
        ├── 05_training_mechanics_and_eval.ipynb
        └── 06_hackathon_template.ipynb
    └── workbooks/
        ├── 00_intro_and_data_exploration.ipynb
        ├── 01_linear_regression_medical.ipynb
        ├── 02_regularization_high_dim_medical.ipynb
        ├── 03_feature_engineering_and_nonlinearity.ipynb
        ├── 04_mlp_for_tabular_medical.ipynb
        ├── 05_training_mechanics_and_eval.ipynb
        └── 06_hackathon_template.ipynb

Installation and Setup

1. Clone or Download

git clone https://github.com/QLS-MiCM/SupervisedML.git
cd SupervisedML

2. Create Environment

Using conda:

conda create -n micm-ml python=3.10 -y
conda activate micm-ml

Using venv:

python -m venv micm-ml
source micm-ml/bin/activate # Windows: micm-ml\Scripts\activate

3. Install Dependencies

pip install -r requirements.txt

4. Generate Datasets

cd Exercises/scripts
python generate_datasets.py

This creates five datasets inside Exercises/data/.

5. Launch Jupyter

jupyter lab

or

jupyter notebook

Then open the notebooks in Exercises/workbooks/.

6. To generate slides (optional)

Only needed if you edit the slide markdown in Slides/markdown/.

Requirements:

• Node.js LTS (v22 recommended; v26 is not supported by the pinned marp-cli version)
• Google Chrome, Microsoft Edge or Firefox

npm install            # installs marp-cli from package.json
bash build_slides.sh   # rebuilds Slides/markdown/*.md → Slides/PDFs/*.pdf

On Windows, if marp cannot find a browser, point it to Edge first:

export CHROME_PATH="/c/Program Files (x86)/Microsoft/Edge/Application/msedge.exe"

Dependencies

See requirements.txt for exact versions. Core dependencies include:

• numpy
• pandas
• scikit-learn
• matplotlib
• jupyter
• statsmodels

License

This material is for educational purposes within the McGill Initiative in Computational Medicine (MiCM), released under MIT license.

Generative AI declaration

Portions of this workshop were drafted with assistance from Generative AI tools (ChatGPT). All content has been reviewed, edited, and validated by the instructor to ensure accuracy and clarity.

Workshop created as part of the McGill Initiative in Computational Medicine