Work with data
Use Python, Colab, data handling, plots, and AI-assisted debugging with confidence.
To learn Full schedule
Teaching and assessment
Lecture Lab Workshop Assessment
Week Week 1
Lecture
Introduction of machine learning and tools Basic
Learning outcomes
Core · all students
- Set up Google Colab and run basic Python code
- Save and load data
- Create simple plots using matplotlib
- Simple data exploration and cleaning
- Use AI tools (ChatGPT/Copilot) to debug code
Advanced · MPS439
- Explore advanced IDE (VS Code/Cursor)
Assessment —
Week Week 1
Lab
Tool usage: iPython notebook, Google Colab
Programming: Python basics and visualization
Assessment —
Week Week 2
Lecture
linear regression (with a machine learning framework, input, output, loss, training) regression
Learning outcomes
Core · all students
- Apply sklearn's LinearRegression to predict continuous outcomes
- Explain why minimizing squared error makes sense
- Understand metrics like R2, MSE, MAE, etc
- Describe what the normal equation solution tells us
- Compare training vs testing performance
Advanced · MPS439
- Reproduce full least squares derivation
- Implement gradient descent from scratch
Assessment Assessment 1 Release
Week Week 2
Lab
linear regression
Assessment —
Week Week 3
Lecture
linear regression Plus
Learning outcomes
Core · all students
- Apply simple feature engineering
- Use Ridge and Lasso regression in sklearn
- Explain why Ridge regression shrinks coefficients toward zero
- Describe when regularization prevents overfitting
- Choose regularization strength using cross-validation
Advanced · MPS439
- Derive Ridge/Lasso cost functions using Lagrange multipliers
- Implement regularized regression from scratch
Assessment —
Week Week 3
Lab
linear regression Plus
Assessment —
Week Week 4
Lecture
linear classification (with metrics for classifications) classification
Learning outcomes
Core · all students
- Apply LogisticRegression for binary classification
- Explain why we use sigmoid function for probabilities
- Describe when linear classification fails
- Interpret confusion matrices and classification metrics, ROC curve, AUC
Advanced · MPS439
- Derive cross-entropy loss function
- Implement logistic regression using gradient descent
Assessment Exercise 1 (optional)
Release
Week Week 4
Lab
linear classification
Assessment —
Week Week 5
Lecture
Discriminative analysis
Learning outcomes
Core · all students
- Use sklearn's LinearDiscriminantAnalysis and QuadraticDiscriminantAnalysis
- Explain why LDA assumes equal covariances
- Describe when LDA vs QDA is appropriate
- Compare LDA vs logistic regression performance
Advanced · MPS439
- Derive LDA decision boundaries mathematically
- Implement LDA in python
Assessment —
Week Week 5
Workshop (Optional)
Questions on assignments; General discussions;
Demonstration of AI for study
Assessment —
Week Week 5
Lab
Discriminative analysis
Assessment —
Week Week 6
Lecture
Decision Tree
Learning outcomes
Core · all students
- Build decision trees using sklearn's DecisionTreeClassifier
- Explain how splitting criteria (Gini, entropy) work
- Describe when trees overfit and how to prevent it
- Interpret tree structure and feature importance
Advanced · MPS439
- Calculate Gini impurity by hand
- Explore using other classification method in sklearn (eg, SVM)
- Explore advanced ensemble methods (XGBoost)
Assessment Assessment 1 submission deadline
Week Week 6
Lab
Decision Tree
Assessment —
Week Week 7
Lecture
Reading week
Assessment Assessment 2 Release
Week Week 8
Lecture
Principal component analysis (PCA) Dimension reduction (non-supervise learning)
Learning outcomes
Core · all students
- Use sklearn's PCA for dimensionality reduction
- Explain why PCA finds directions of maximum variance
- Describe what eigenvalues represent
- Determine optimal number of components
Advanced · MPS439
- Derive PCA from eigenvalue decomposition
- Implement PCA using SVD from scratch
Assessment Exercise 1 (optional) submission deadline
Week Week 8
Lab
Principal component analysis (PCA)
Assessment —
Week Week 9
Lecture
k-means and Hierarchical Clustering Clustering (non-supervise learning)
Learning outcomes
Core · all students
- Apply K-means clustering using sklearn
- Explain why K-means finds spherical clusters
- Describe when hierarchical vs K-means is better
- Choose optimal K using elbow method
Advanced · MPS439
- Implement K-means algorithm from scratch
Assessment —
Week Week 9
Lab
k-means and Hierarchical Clustering
Assessment —
Week Week 10
Lecture
Deep learning Deep Learning
Learning outcomes
Core · all students
- Implementing neural networks using Keras
- Explain why we need multiple layers
Advanced · MPS439
- Understand the backpropagation algorithm
- Implement a simple neural network with different tricks: regularization, dropout, stochastic gradient descent, etc
Assessment Assessment 1 feedback to student
Week Week 10
Lab
Deep learning
Assessment —
Week Week 11
Lecture
Deep learning Convolutional network
Learning outcomes
Core · all students
- Understand how a convolutional layer works
- Implementing a CNN using Keras
Advanced · MPS439
- Build Convolutional neural networks using Keras
- Describe what convolutions detect in images
- Load pre-built CNN (eg, vgg-16) or other architecture to a classification problem
Assessment Exercise 1 (optional) feedback to student
Week Week 11
Workshop (Optional)
Questions on assignments; General discussions;
Demonstration of AI for study; convolutional networks and applications
Assessment —
Week Week 11
Lab
Deep learning Convolutional network
Assessment —
Week Week 13
Assessment
Assessment 2 Submission deadline
Assessment —
Week —
Assessment
Assessment 2 feedback to student
Assessment —