Machine Learning Intro Practice

This repository is a step-by-step introduction to practical data science and entry-level machine learning for young research students.

The example project uses a sample materials-science dataset about twin-boundary migration. The dataset is synthetic, so it is safe for practice: students can learn the workflow before using real research data.

Who This Is For

This repo is designed for students who:

• know basic Python syntax, or are learning it;
• have some statistics background;
• are new to pandas, plotting, Jupyter notebooks, and scikit-learn;
• need a practical path from setup to a small interpretable ML result.

Learning Goals

By the end, students should be able to:

• explain what data science does;
• set up a clean Python environment;
• use Jupyter notebooks for research notes and code;
• load and inspect tabular data with pandas;
• make clear exploratory plots;
• define features and a target;
• compare a baseline model with simple ML models;
• evaluate models with MAE, RMSE, and R2;
• make a feature-importance plot;
• explain what a model can and cannot prove.

Start Here

1. Read docs/project_overview.md.
2. Set up Python using one guide:
   • Windows: docs/setup_windows.md
   • Mac: docs/setup_mac.md
3. Follow the main practice checklist: exercises/checklist.md.
4. Use the lessons in order:
   • lessons/01_environment_check.md
   • lessons/02_data_science_workflow.md
   • lessons/03_pandas_toy_data.md
   • lessons/04_visualization_eda.md
   • lessons/05_baseline_and_regression.md
   • lessons/06_random_forest_interpretability.md
   • lessons/07_mini_project.md

For instructor publishing notes, see docs/publishing_to_github.md.

Suggested Time

The first pass is designed for 6-8 hours:

Block	Time
Project overview and reflection	30-45 min
Python/Jupyter setup	60-90 min
Notebook smoke test	20-30 min
pandas and toy dataset	60-90 min
Exploratory plots	90-120 min
Baseline and regression	90-120 min
Random forest and feature importance	60-90 min
Research log	30-45 min

Repository Layout

ml_intro_practice/
  README.md
  requirements.txt
  docs/
    project_overview.md
    setup_windows.md
    setup_mac.md
    reading_list.md
  lessons/
    01_environment_check.md
    02_data_science_workflow.md
    03_pandas_toy_data.md
    04_visualization_eda.md
    05_baseline_and_regression.md
    06_random_forest_interpretability.md
    07_mini_project.md
  exercises/
    checklist.md
  scripts/
    generate_toy_data.py
    run_model_demo.py
  data/
    README.md
  notebooks/
    README.md

Quick Setup

After creating and activating a virtual environment:

python -m pip install --upgrade pip
python -m pip install -r requirements.txt
jupyter lab

For full setup instructions, use the operating-system guides in docs/.

Instructor Notes

This repo is intentionally modest. It teaches the core habits before advanced models:

• start with the question;
• understand the table;
• plot before modeling;
• compare against a baseline;
• evaluate on held-out data;
• interpret carefully;
• write down limitations.

The toy dataset is generated by scripts/generate_toy_data.py. The numbers are not physics. They are designed to create a small, interpretable tabular ML exercise.