Fundamentals of Machine Learning

📋 Task 1: Exploratory Data Analysis (15 points)

• Perform exploratory data analysis on your assigned dataset
• Create meaningful visualizations to understand data distributions and relationships
• Formulate and test hypotheses based on your observations
• 🔑 MANDATORY: Comment your analysis insights in Markdown cells
• Explain the patterns, outliers, and correlations you discover

🎯 Task 2: Clustering Analysis (30 points)

Part 1: K-Means Clustering

• Apply K-Means clustering algorithm to your preprocessed data
• 🚀 Use the elbow method and/or silhouette coefficient to determine optimal cluster count
• Visualize results using boxplots, bar plots, and scatter plots
• 🔑 MANDATORY: Define cluster characteristics in Markdown cells
• Compare feature values across different clusters

Part 2: Alternative Clustering Algorithm

• Choose another algorithm from sklearn documentation
• 🚀 Agglomerative clustering is recommended as a good starting point
• Research the algorithm through blog posts, books, or videos
• Repeat the analysis process from Part 1
• Compare results with K-Means clustering

📋 Data preparation

• Every project must include description of the dataset features and a brief exploration analysis so you get to know the data you are working with
• 📒Describe what operations you are performing for each of the features
  • e.g. No operation needed, dropping the feature, Min-Max scaling, One-hot encoding, Categorization (numeric -> categorical) etc.
  • You can even try to experiment with many different preprocession operations and you will see what works the best - this is totally up to you!
• 🔎 What model performance metrics have you decided to use?
  • e.g. Accuracy, Precision, Recall, F1-score etc.
  • 📒 State which one of the scores is the most important from your point of view given the class balance, task, ... in a Markdown cell!

🎯 Additional requirements

• 🎯 Try at least 3 different classification algorithms
  • 🌳 Decision tree is a mandatory one
  • Other algorithm choices depends solely on you - you can find many different algorithms in the scikit-learn docs.
  • We can recommend the Multi-layer Perceptron (MLP), Random Forest, AdaBoost, k-Nearest Neighbors or Naive Bayes.
  • Check blog posts, book chapters or Youtube videos about the algoritmhs you have chosen so you have a basic knowledge about its key concepts.
  • 💡 You know some of them already from the lectures 🙂
• k-Fold cross validation will be used for the algoritmh performance evaluation
  • e.g. 5-fold, 10-fold, ...
  • 📒 State if you used standard or stratified version and why in a Markdown cell.
• 🚀 Your project must include hyper-parameters tuning
  • Given the fundamental knowledge about the selected algorithms select suitable parameters for the tuning (See scikit-learn docs for details).
  • Do at least 15 experiments with different values of selected parameters in total.
  • 💡 This part can be imbalanced! You don't have to try 5 different setting for each of algorithm - You can focus on a single algorithm more however at least 2 parameter settings have to be tried for every algorithm e.g. 2 for Decision Tree, 2 for Random Forest, 11 for MLP
  • 📊 Create a table with header like | Algorithm | Parameters | Metrics score | for you experiments results
  • 📒 Explain the parameters effect on the model performance in a Markdown cell