Data Mining and Knowledge Discovery Course

Lesson 1 • What Is Data Mining

Defines data mining, its objectives, and its role within the broader KDD process. Grounds all subsequent techniques in a shared conceptual framework.

Lesson 2 • Core Data Mining Tasks

Introduces classification, regression, clustering, and association rule mining as the primary task families. Links each task to real-world problem types students will encounter.

Lesson 3 • Ethical and Privacy Considerations

Covers bias, fairness, consent, and data governance principles relevant to mining projects. Establishes responsible practice standards applied throughout the course.

Lesson 1 • Data Sampling and Imbalance Handling

Explains random, stratified, and cluster sampling strategies alongside techniques for imbalanced class distributions. Ensures representative training sets for reliable model evaluation.

Lesson 2 • Data Integration and Transformation

Covers merging heterogeneous sources, resolving schema conflicts, and applying normalisation and discretisation. Ensures data from multiple origins is consistent and algorithm-ready.

Lesson 3 • Data Cleaning Techniques

Addresses missing values, outliers, noise, and inconsistencies that degrade model performance. Builds the practical cleaning skills required before any algorithm is applied.

Lesson 1 • Univariate and Bivariate Visualisation

Covers histograms, box plots, scatter plots, and heatmaps for single and paired variable analysis. Equips students to detect distributions, outliers, and relationships visually.

Lesson 2 • Descriptive Statistics for Mining

Reviews central tendency, dispersion, skewness, and correlation as diagnostic tools for data understanding. Connects statistical summaries to decisions about preprocessing and algorithm choice.

Lesson 3 • Dashboard Design for Data Insights

Teaches layout principles, interactivity, and storytelling for analytical dashboards. Prepares students to present EDA findings to both technical and non-technical stakeholders.

Lesson 1 • Instance-Based and Kernel Methods

Introduces k-nearest neighbours and support vector machines as non-parametric classifiers. Connects kernel tricks and distance metrics to practical performance trade-offs.

Lesson 2 • Decision Trees and Rule Learners

Explains splitting criteria, tree pruning, and rule extraction for interpretable classification. Provides the conceptual base for understanding more complex ensemble methods.

Lesson 3 • Ensemble Methods

Teaches bagging, boosting, and stacking to combine weak learners into strong predictors. Demonstrates how ensembles reduce variance and bias beyond single-model limits.

Lesson 4 • Classifier Evaluation and Model Selection

Covers confusion matrices, ROC curves, cross-validation, and statistical tests for model comparison. Ensures students can objectively choose the best classifier for a given task.

Lesson 1 • Regularisation Methods

Introduces Ridge, Lasso, and Elastic Net penalties to control overfitting in high-dimensional settings. Connects regularisation strength to feature selection and bias-variance trade-off.

Lesson 2 • Nonlinear and Tree-Based Regression

Extends prediction to polynomial, spline, and tree-based models for capturing complex relationships. Demonstrates when nonlinear models outperform linear baselines.

Lesson 3 • Linear and Multiple Regression

Covers ordinary least squares estimation, coefficient interpretation, and assumption checking. Establishes the regression baseline against which all advanced methods are compared.

Lesson 1 • Density-based and grid-based clustering

Introduces DBSCAN and OPTICS for discovering arbitrarily shaped clusters and handling noise. Contrasts density approaches with centroid methods for complex data distributions.

Lesson 2 • Hierarchical clustering approaches

Explains agglomerative and divisive strategies, linkage criteria, and dendrogram interpretation. Enables students to cluster without specifying k in advance.

Lesson 3 • Partitional clustering methods

Covers k-means and k-medoids algorithms, initialisation strategies, and convergence behaviour. Establishes the foundational clustering paradigm before introducing more complex variants.

Lesson 1 • Efficient mining algorithms

Covers the FP-Growth algorithm and its compressed tree structure as a scalable alternative to Apriori. Demonstrates computational advantages on large transactional datasets.

Lesson 2 • Rule evaluation and pruning

Introduces interestingness measures beyond confidence to filter redundant and spurious rules. Equips students to deliver concise, high-value rule sets to stakeholders.

Lesson 3 • Frequent itemset mining fundamentals

Defines support, confidence, and lift and explains the Apriori principle for pruning the search space. Provides the theoretical basis for all association rule algorithms.

Lesson 1 • Mining at scale with big data tools

Introduces distributed computing frameworks and in-memory processing for large-scale mining tasks. Prepares students to apply core algorithms on datasets exceeding single-machine capacity.

Lesson 2 • Model deployment and monitoring

Teaches model serialisation, REST API serving, and production monitoring for concept drift. Completes the KDD pipeline from raw data to a maintained, live prediction service.

Lesson 3 • Neural networks for data mining

Covers feedforward networks, backpropagation, and activation functions applied to classification and regression. Bridges classical mining algorithms to modern deep learning approaches.

Lesson 4 • End-to-end capstone project

Integrates all course skills into a complete mining project from problem definition to deployed model. Develops professional portfolio artefacts demonstrating full KDD competency.