UUY-DTS 3132 units300 LevelFirst Semester

Machine Learning and Applications

B.Sc. Data Science, University of Uyo

The core machine learning course: supervised and unsupervised algorithms, how they are trained, how they are evaluated, and how they fail. The technical centrepiece of the degree.

On Areté, UUY-DTS 313 comes with

  • 3 recommended textbooks
  • an AI tutor that has read this course’s outline and notes
Sign in to study UUY-DTS 313

Free for University of Uyo students — sign in with your email, no password.

Course outline

  1. 01The learning problem: supervised, unsupervised, semi-supervised and reinforcement learning
  2. 02Linear and logistic regression; loss functions and gradient descent
  3. 03Regularisation: ridge, lasso, and the bias–variance trade-off revisited
  4. 04k-nearest neighbours, Naive Bayes and support vector machines
  5. 05Decision trees, and ensembles: bagging, random forests and gradient boosting
  6. 06Unsupervised learning: k-means, hierarchical clustering, DBSCAN
  7. 07Dimensionality reduction: principal component analysis and t-SNE
  8. 08Model evaluation: confusion matrix, precision, recall, F1, ROC/AUC, cross-validation, hyperparameter tuning
  9. 09Introduction to neural networks: perceptrons, backpropagation, and where deep learning applies

Recommended textbooks

  • An Introduction to Statistical Learning — James, Witten, Hastie & Tibshirani

    2nd ed. — free at statlearning.com; the ideal level for this course

  • Hands-On Machine Learning with Scikit-Learn, Keras and TensorFlow — Aurélien Géron

    3rd ed. — the best practical companion

  • The Elements of Statistical Learning — Hastie, Tibshirani & Friedman

    2nd ed. — free; the rigorous reference for when you want the mathematics

How to pass UUY-DTS 313

  • Accuracy is a misleading metric on imbalanced data — learn precision, recall and the ROC curve properly, and be able to argue which matters for a given problem
  • Implement linear regression and k-means from scratch in NumPy once. After that, using scikit-learn is informed rather than magical
  • For every algorithm, learn its inductive bias — what it assumes about the data. That is what tells you when to reach for it
  • Compete in one beginner Kaggle competition; the feedback loop teaches evaluation discipline faster than any lecture

Other 300 Level courses

Lecture notes, practice questions and the AI tutor for UUY-DTS 313 are available once you sign in.

Sign in to Areté