Advanced Methods in Machine Learning Applications

Projects You’ll Build (With Expert Guidance)

With ready-to-use Jupyter notebooks and working code examples, Dr. Guven will walk you through creating…

• Titanic Ensemble Classifier
  Revisit the classic Titanic dataset to compare baseline classifiers with ensemble methods. You’ll build a Random Forest ensemble using decision trees trained on random feature subsets, then combine them via majority voting—illustrating how ensembles can surpass individual models, even when those models are intentionally limited.
• Heart Disease Ensemble Benchmark
  Using a real-world heart disease dataset, you’ll build ensemble classifiers from decision trees, neural networks, SVMs, and Naive Bayes models. You’ll benchmark their performance under subsampled training conditions and compare how well each ensemble generalizes, even when individual models are handicapped.
• Cancer Recurrence Predictor (Advanced Version)
  Predict cancer recurrence using logistic regression and a fully preprocessed medical dataset. You’ll apply one-hot encoding, scale features, and evaluate the model using accuracy and feature ranking—demonstrating how interpretable models support transparent decision-making.
• Advanced Regression Explorer
  Fit and compare linear, polynomial, and logistic regression models on real datasets, including curve-fitting scenarios. You’ll use metrics like MAE, MSE, and MAPE to evaluate model fit—and explore why linear regression sometimes fails, even when the metrics say it shouldn’t.
• Customer Behavior Analysis with Apriori
  Apply Apriori analysis to uncover hidden relationships in transactional data, critical for building effective recommender systems.
• Pattern Discovery via Clustering (Iris dataset)
  Explore unsupervised learning by clustering the classic 1936 Iris flower dataset, discovering natural groupings without relying on labeled outcomes.