Machine Learning Tutorials (Jupyter Notebooks) — English & فارسی

A bilingual, notebook-first machine learning course delivered in Jupyter Notebook (.ipynb) format.
Each lesson is provided in:

• English: Tutorials/English/...
• Persian (Farsi / فارسی): Tutorials/Persian/... (files end with _Fa.ipynb)

The English and Persian notebooks cover the same material, with the Persian version being a translation of the English content.

---

Repository layout

Datasets/
  Classification/   # tabular datasets for classification examples
  Regression/       # tabular datasets for regression examples
  Clustering/       # datasets for unsupervised learning and clustering
Tutorials/
  English/
    Chapter1/
      Chapter1_Lesson1.ipynb
      Chapter1_Lesson2.ipynb
      ...
    ...
    Chapter37/
      Chapter37_Lesson1.ipynb
      Chapter37_Lesson2.ipynb
      ...
  Persian/
    Chapter1/
      Chapter1_Lesson1_Fa.ipynb
      Chapter1_Lesson2_Fa.ipynb
      ...
    ...
    Chapter37/
      Chapter37_Lesson1_Fa.ipynb
      Chapter37_Lesson2_Fa.ipynb
      ...
css/
  rtl.css           # RTL styles used by Persian notebooks

• Datasets/ contains CSV files used in examples, exercises, and projects.
• Tutorials/ contains the notebooks organized by Chapter → Lesson.
• css/rtl.css supports right-to-left rendering (Persian) and related typography.

---

What you will learn (high-level)

This course spans an end-to-end classical machine learning curriculum, progressing from fundamentals to advanced topics and applied case studies:

• Foundations: what ML is, problem framing, learning paradigms, workflow, and reproducibility concepts.
• Data & preprocessing: data types, cleaning, missing data, encoding, scaling, feature engineering, leakage prevention, and pipeline hygiene.
• EDA and diagnostics: visualization, correlation and relationships, hypothesis testing in ML context, reporting, and practical data-quality forensics.
• Supervised learning: regression and classification, linear/logistic models, regularization, GLMs, multiclass/ordinal settings, robust methods, and cost-sensitive decisioning.
• Tree-based learning: CART and related algorithms, pruning, interpretability, feature importance, and practical concerns (missing/categorical handling, stability).
• Ensembles: bagging, boosting, stacking, calibration, imbalance handling, and modern gradient boosting families (including advanced tuning and interpretability ideas).
• Kernel & instance-based methods: kNN, distance metrics, curse of dimensionality, metric learning overview, and kernel methods beyond SVM.
• Probabilistic modeling: Naïve Bayes variants, Bayesian perspectives, priors/MAP, and EM intuition.
• Unsupervised learning: clustering families (k-means, hierarchical, mixtures, density-based, spectral) plus evaluation and stability concepts.
• Dimensionality reduction: PCA, t-SNE, LDA, ICA, NMF, random projections, and embedded feature selection.
• Model evaluation: cross-validation strategies, calibration, learning curves, nested CV, uncertainty in metrics, and decision analysis.
• Time series: classical forecasting workflows, ARIMA/Prophet concepts, proper CV, feature engineering, and anomaly/event detection fundamentals.
• Reliability & uncertainty: calibration, prediction intervals, conformal prediction basics, and shift-aware evaluation.
• Responsible ML: fairness, privacy, robustness, and security considerations for classical ML systems.
• MLOps (classical): experiment tracking concepts, versioning, deployment patterns, monitoring, governance, and maintenance practices.
• Capstone projects: end-to-end, notebook-driven projects across classification, regression with uncertainty, clustering/segmentation, forecasting, anomaly detection, and causal/uplift thinking.

---

Getting started

1) Install Python (recommended)

• Install Python 3.9+ (or a newer 3.x release).
• Recommended option for data science workflows: Anaconda / Miniconda.

2) Install Jupyter (Windows / macOS / Linux)

Option A — Install via pip (works on all platforms)

python -m pip install --upgrade pip
python -m pip install notebook jupyterlab

Then launch:

jupyter lab
# or
jupyter notebook

Option B — Install via conda (recommended if you use Anaconda/Miniconda)

conda install -c conda-forge notebook jupyterlab

Then launch:

jupyter lab
# or
jupyter notebook

3) Install course dependencies (requrement.txt)

This repository includes a requrement.txt file (intentionally kept as the repo’s dependency manifest).
It lists the Python packages used throughout the notebooks (NumPy, Pandas, Matplotlib, SciPy, scikit-learn, and supporting tools).

Option A — Create a virtual environment (recommended) and install with pip

From the repository root:

# Create a virtual environment
python -m venv .venv

# Activate it (Windows PowerShell)
.venv\Scripts\Activate.ps1

# Activate it (Windows CMD)
.venv\Scripts\activate.bat

# Activate it (macOS/Linux)
source .venv/bin/activate

# Upgrade pip and install dependencies
python -m pip install --upgrade pip
python -m pip install -r requrement.txt

Then start Jupyter from the repo root:

jupyter lab
# or
jupyter notebook

Option B — Use conda and install with pip

conda create -n ml-tutorials python=3.11
conda activate ml-tutorials
python -m pip install --upgrade pip
python -m pip install -r requrement.txt

---

Download / clone the repository

Download as ZIP

1. Click Code → Download ZIP on GitHub.
2. Extract the ZIP file to a local folder.

Clone with Git

git clone https://github.com/<YOUR-USERNAME>/<YOUR-REPO>.git
cd <YOUR-REPO>

---

Open notebooks so only this repo appears

To ensure Jupyter shows only the files/folders inside this repository, start Jupyter from the repo root directory.

JupyterLab (recommended)

cd <YOUR-REPO>
jupyter lab

Jupyter Notebook (classic)

cd <YOUR-REPO>
jupyter notebook

(Optional) Force notebook root explicitly

If you want to be explicit about the working directory:

cd <YOUR-REPO>
jupyter notebook --notebook-dir="."

From the file browser, navigate to:

• Tutorials/English/ChapterX/ for English notebooks
• Tutorials/Persian/ChapterX/ for Persian notebooks (*_Fa.ipynb)

---

Important: run the first cell first (CSS + RTL + fonts)

In each notebook, the first cell is responsible for applying display styling (for example, loading CSS for right-to-left layout and fonts for Persian).

Before reading or executing anything else:

1. Open the notebook
2. Run the first cell
3. Continue with the remaining lesson cells

This ensures the Persian notebooks render correctly and consistently.

---

YouTube walkthrough

A short video walkthrough shows how to download/clone the repo and run the notebooks.