Machine Learning 2 – Exercise Solutions

Author: Taha El Amine Kassabi
Course: Machine Learning 2 (WS 2023/24)
Instructor: Prof. Dr. Dominik Heider
University: Heinrich Heine University Düsseldorf (HHU)

---

📚 Overview

This repository contains solutions to all 11 exercises from the Machine Learning 2 course at HHU. Each assignment covers foundational or advanced topics in classical machine learning — from matrix manipulation and statistics to support vector machines and neural networks. Most solutions consist of Jupyter notebooks, supplementary markdown or PDF answers, and linked datasets.

---

📂 Repository Structure

Sheets/
├── Sheet 0 – Conda, YAML, file IO
├── Sheet 1 – Text processing & matrices
├── Sheet 2 – Descriptive statistics, correlation
├── Sheet 3 – PCA & MDS (PCoA)
├── Sheet 4 – Clustering (K-Means, Hierarchical)
├── Sheet 5 – Linear and logistic regression
├── Sheet 6 – k-Nearest Neighbors
├── Sheet 7 – Support Vector Machines
├── Sheet 8 – Decision Trees
├── Sheet 9 – Random Forest & Performance Measures
└── Sheet 10 – Neural Nets, Backprop, CNN filters

Each folder includes a task sheet (exercise_N.pdf) and data or helper files. Solutions are under Solutions/Exercise N.

---

🧠 Topics Covered

Exercise	Topics
0	Conda setup, YAML, reading/writing files
1	Text parsing, matrix algebra
2	Descriptive stats, skewness, correlation
3	PCA & classical MDS
4	K-means, silhouettes, hierarchical clustering
5	Linear regression, logistic regression
6	k-Nearest Neighbors (k-NN), normalization
7	SVM, Lagrange multipliers, separating hyperplanes
8	Decision trees, information gain, Gini
9	Random forests, Gini importance, AUC, precision/recall
10	McCulloch/Pitts networks, backpropagation, CNN filters, MNIST training (bonus)

---

💾 Setup

pip install -r requirements.txt

You may also use JupyterLab or Google Colab for individual notebooks.

---

🚀 Usage

Navigate to any exercise in Solutions/Exercise N, open the notebook(s), and run the cells. Some tasks also include .md, .pdf, or .docx explanations for handwritten answers.

---

📝 Notes

• Most exercises were completed using Python with NumPy, Pandas, Matplotlib, Scikit-learn, and **SciPy **.
• For neural networks (Exercise 10 bonus), models were implemented using Keras/TensorFlow and PyTorch.