Machine Learning

📘 Machine Learning

This course provided a comprehensive introduction to modern ML workflows: from data preparation and feature engineering to the evaluation of supervised and unsupervised models.
We covered classification (SVM, decision trees, random forests, naïve Bayes), regression (linear, RANSAC, Theil-Sen, Huber, ensembles), clustering (K-Means, DBSCAN, Gaussian Mixtures), and model interpretability techniques.
The methodology strongly combined theory and practice, including hands-on projects and model optimization.

👉 Official Teaching Guide (Deusto)

---

🔗 Related Projects & Applications

• Health Risk Prediction (Classification: Smoking/Drinking)
  Repo: Smoking-Drinking-analysis
  Applied multiple classifiers (Decision Trees, Random Forests, Naïve Bayes, SVM, SGD, AdaBoost, LightGBM, XGBoost) with Optuna hyperparameter tuning to predict lifestyle risk factors.

• Craigslist Used Cars (Regression)
  Repo: Used-Cars-Price-Kaggle
  Implemented and optimized regression models (Linear, Random Forest, Polynomial, SVM, SGD) and ensemble methods with Optuna, achieving strong predictive performance on ~200k car listings.

• Mall Customer Segmentation (Clustering)
  Repo: Mall-Customer-Segmentation-Data-Kaggle
  Performed unsupervised clustering (K-Means, DBSCAN, Gaussian Mixtures, Agglomerative, OPTICS, BIRCH), combined with PCA for visualization. Compared clustering approaches to optimize segmentation strategies.

---

📌 This course was pivotal for consolidating my skills in ML: not only implementing algorithms but also performing critical evaluation, tuning, and applying them to real-world datasets.