CERN School of Computing - Latin America 2026

Notebooks covered in the lecture "Data science and data exploration"

🚀 Getting started

These exercises are designed to run on the CERN SWAN service. You will need an active CERN account to proceed.

---

1. Access SWAN

Go to https://swan.cern.ch and log in using your CERN credentials.

---

2. Configure your SWAN session

When the session setup page appears:

1. Select 108 in the Software stack dropdown.
2. Ensure “Use Python packages installed on CERNBOX” is checked.
3. Click “Start my Session” (bottom right).
4. Wait for the environment to launch — this may take about one minute.

---

3. Download the project repository

When you reach the “My Projects” screen in SWAN:

1. Click “Download project from git” (the small cloud-and-arrow icon in the upper-right corner)
   

2. Enter the repository URL:

https://github.com/lserkin/csc2026-lecture.git  

3. Click “Download” once.

Please wait for the download to complete — this may take a couple of minutes.

---

4. Running the notebooks

Once the download is complete, you will see yourself within a directory named csc2026-lecture.

Pick the notebook (.ipynb file) you want and launch it (click), a new window will be opened.

Run the notebook by clicking the “▶” Run button repeatedly until all cells have finished executing.

You can also run everything by going into "CELL" and then "Run All"

Instructions are embedded in each file. For some exercises, there are also data files that you can browse by opening them in SWAN.

---

---

General information

If you want to run these on your own machine, first install Jupyter Notebooks. Then checkout the exercise material to your machine. You can also run these notebook in Google Colab.

Documentation

Core scientific Python stack

• numpy package
  Numerical computing, arrays, linear algebra, random sampling.

• pandas package
  DataFrames, I/O tools, data cleaning, grouping & time series.

  • API reference

• matplotlib package
  Low-level plotting library used for scientific and publication-quality figures.

Additional libraries

• seaborn package
  High-level statistical visualization built on top of Matplotlib.

• plotly package
  Interactive plotting library for dynamic graphs, dashboards, and 3D visualization.

• scikit-learn package
  Machine learning tools for:

  • preprocessing and feature engineering
  • dimensionality reduction (PCA, t-SNE, UMAP)
  • clustering (KMeans, DBSCAN, Agglomerative)
  • modeling (regression, classification)
  • model evaluation and pipelines
  • synthetic dataset generation (make_moons, make_blobs, etc.)