Practical Notebooks — ML Foundations

This repository contains four hands-on notebooks used for course tutorials (CIS6530).
Each notebook is designed to be easy to run, self-contained, and includes evaluation + visualizations.

Contents

1) Supervised Learning (Classification)

Notebook: supervised_learning_example.ipynb

• Dataset: Iris (built-in via scikit-learn)
• Task: multi-class classification
• Model: Logistic Regression, k-NN
• Evaluation: confusion matrix, accuracy, precision/recall/F1

2) Unsupervised Learning (Clustering)

Notebook: unsupervised_learning_example.ipynb

• Dataset: Iris (features only; labels used only for optional comparison)
• Methods: k-means, DBSCAN, Agglomerative clustering
• Evaluation: silhouette score, plots in PCA space

3) Deep Learning

Notebook: deep_learning_example.ipynb

• Dataset: KDD Cup ’99
• Task: intrusion detection (NORMAL vs ATTACK)
• Model: MLP (feed-forward neural network)
• Evaluation: confusion matrix, classification report, error inspection, threshold tuning

4) LLM-based Methods

Notebook: llm_practical_example.ipynb

• Dataset: Enron spam/ham via Hugging Face Datasets (SetFit/enron_spam)
• Model: google/flan-t5-small (runs locally on CPU)
• Task: spam detection (SPAM vs HAM) using prompting
• Evaluation: confusion matrix, classification report, error analysis, few-shot prompting