notebooks

Notebooks

Runnable Jupyter notebooks. Each one is self-contained — open it, run it top to bottom, then change something and see what breaks.

← Back to the README

---

What's here

Notebook	Topic	Level	Contributor
knn-iris.ipynb	K-Nearest Neighbours with full EDA on Iris	Beginner	MOSS AI Chapter
random-forests.ipynb	Random Forests — compact walkthrough	Beginner	MOSS AI Chapter
random-forests-deep-dive.ipynb	Random Forests in depth, with diagrams	Intermediate	Keshav Krishna Singh
gradient-boosting.ipynb	XGBoost, CatBoost and LightGBM compared	Intermediate	MOSS AI Chapter
perceptron.ipynb	The perceptron built from scratch in TensorFlow	Intermediate	Hrishik Patel

Supporting images live in assets/.

---

Running them

From the repository root:

python -m venv .venv
source .venv/bin/activate          # Windows: .venv\Scripts\activate
pip install -r requirements.txt

cd notebooks
jupyter notebook

Or open any notebook in Google Colab — File → Open notebook → GitHub, then paste AkCodes23/MOSS-AI. On Colab you'll need to install the gradient-boosting libraries yourself:

!pip install xgboost lightgbm catboost

---

Suggested order

1. knn-iris.ipynb — the gentlest start. Covers loading data, cleaning, visualising, and a first classifier, all on a dataset small enough to hold in your head.
2. random-forests.ipynb — your first ensemble method, and why combining weak models beats tuning one.
3. random-forests-deep-dive.ipynb — the same algorithm with the theory filled in: bagging, feature importance, out-of-bag error.
4. gradient-boosting.ipynb — boosting instead of bagging, and a head-to-head of the three libraries that win most Kaggle competitions.
5. perceptron.ipynb — the jump to neural networks, one neuron at a time. From here, go to code/transformers/.

---

Data

Notebook	Data it needs	Where it comes from
knn-iris.ipynb	../data/iris.csv	In this repository
random-forests.ipynb	Iris	Loaded from sklearn.datasets at runtime
random-forests-deep-dive.ipynb	Loaded in-notebook	Included
perceptron.ipynb	None — synthetic	Generated in-notebook
gradient-boosting.ipynb	train.csv, test.csv, sample_submission.csv	Not included — see below

Known gap: gradient-boosting.ipynb reads Kaggle competition files (train.csv, test.csv, sample_submission.csv) for a road-accident-risk prediction task. Those files aren't in this repository. Read the notebook for the method and the library comparison; to execute it, download the competition data from Kaggle into this folder first. If you know the exact competition, a PR adding the link to this table would be genuinely useful.

---

Adding a notebook

See CONTRIBUTING.md. The short version:

• Lowercase, hyphen-separated file name that describes the topic, not the author
• Open with a markdown cell: title, one-line objective, your name
• Relative data paths only — pd.read_csv('../data/iris.csv'), never an absolute path from your own machine
• Restart and run all before committing
• Add a row to the table above and to the main README