Advanced Data Science Course

Lesson 1 • NumPy and Array Computing

Introduces vectorised computation with NumPy arrays for efficient numerical operations. Underpins pandas and machine learning library internals covered later.

Lesson 2 • The Data Science Lifecycle

Maps the full pipeline from problem framing to model deployment. Grounds all subsequent chapters in a shared process model.

Lesson 3 • Data Manipulation with Pandas

Teaches DataFrame operations for loading, filtering, and reshaping tabular data. Directly supports the data preparation work in every subsequent chapter.

Lesson 4 • Core Python for Data Science

Covers Python syntax, data structures, and functional patterns essential for data manipulation. Provides the programming baseline required for all subsequent chapters.

Lesson 1 • Matplotlib and Seaborn Mastery

Builds proficiency with the two primary Python visualisation libraries for static charts. Enables customisation of axes, themes, and annotations for professional output.

Lesson 2 • Univariate and Bivariate Analysis

Examines single-variable distributions and pairwise relationships between features. Reveals patterns that inform feature engineering and model selection.

Lesson 3 • Descriptive Statistics Essentials

Quantifies central tendency, spread, and shape of distributions. Provides the statistical vocabulary used throughout modelling and evaluation chapters.

Lesson 4 • Interactive Dashboards with Plotly

Creates interactive, web-ready visualisations using Plotly and Dash. Prepares students to communicate findings to non-technical stakeholders effectively.

Lesson 1 • Numerical Feature Transformations

Applies scaling, normalisation, and mathematical transforms to improve model convergence. Prepares features for distance-based and gradient-based algorithms.

Lesson 2 • Feature Construction and Selection

Creates domain-driven features and removes redundant ones to improve model performance. Introduces filter, wrapper, and embedded selection methods used in modelling chapters.

Lesson 3 • Imputation and Outlier Treatment

Applies statistical and model-based strategies to fill gaps and handle extreme values. Directly affects model stability and generalisation covered in later chapters.

Lesson 4 • Data Quality Assessment

Identifies and quantifies missing values, duplicates, and inconsistencies in raw datasets. Establishes a systematic audit process applied before any modelling step.

Lesson 1 • Linear Regression Fundamentals

Derives the ordinary least squares objective and interprets coefficients statistically. Establishes the baseline model against which all advanced regressors are compared.

Lesson 2 • Tree-Based Regression Models

Implements decision tree, Random Forest, and Gradient Boosting regressors for nonlinear relationships. Introduces ensemble concepts extended further in the classification chapter.

Lesson 3 • Hyperparameter Tuning for Regression

Optimises model hyperparameters using grid search, random search, and Bayesian optimisation. Establishes tuning workflows reused across all supervised learning chapters.

Lesson 4 • Regularised Regression Methods

Applies Ridge, Lasso, and Elastic Net penalties to control overfitting in high-dimensional data. Connects regularisation theory to the feature selection techniques from Chapter 3.

Lesson 1 • Classification Evaluation Metrics

Quantifies classifier performance using confusion matrices, precision, recall, F1, and AUC-ROC. Connects metric choice to business objectives and class imbalance severity.

Lesson 2 • Logistic Regression and Decision Boundaries

Derives the logistic function and interprets log-odds coefficients for binary classification. Serves as the probabilistic baseline for all classifier comparisons.

Lesson 3 • Advanced Classifiers and Ensembles

Implements SVMs, k-NN, Naive Bayes, and ensemble methods including XGBoost and LightGBM. Extends ensemble theory introduced in the regression chapter to classification tasks.

Lesson 4 • Model Interpretability for Classifiers

Explains classifier predictions using SHAP values and permutation importance. Builds the interpretability foundation expanded in the advanced modelling chapter.

Lesson 1 • Nonlinear Dimensionality Reduction

Applies t-SNE and UMAP to reveal nonlinear manifold structure in high-dimensional data. Extends PCA concepts to complex embeddings used in deep learning visualisation.

Lesson 2 • Cluster Evaluation and Selection

Measures cluster quality using internal and external validation metrics without ground-truth labels. Guides the choice of the number of clusters and algorithm selection.

Lesson 3 • Principal Component Analysis

Reduces dimensionality by projecting data onto orthogonal axes of maximum variance. Prepares high-dimensional data for visualisation and downstream supervised models.

Lesson 4 • Clustering Algorithms

Partitions data into meaningful groups using k-means, hierarchical, and density-based methods. Provides segmentation tools applicable to customer analytics and anomaly detection.

Lesson 1 • Recurrent Networks and Sequence Modelling

Models sequential dependencies using RNNs, LSTMs, and GRUs for time series and text data. Prepares students for the NLP and time series chapters that follow.

Lesson 2 • Neural Network Fundamentals

Builds intuition for perceptrons, activation functions, and forward propagation. Establishes the mathematical foundation for all deep learning architectures that follow.

Lesson 3 • Training Deep Networks

Covers gradient descent variants, learning rate schedules, and batch normalisation for stable training. Directly addresses the vanishing gradient and slow convergence problems.

Lesson 4 • Convolutional Neural Networks

Implements CNNs for image classification using convolutional, pooling, and fully connected layers. Introduces transfer learning from pretrained models to reduce training data requirements.

Lesson 1 • Experiment Tracking and Model Registry

Logs hyperparameters, metrics, and artefacts using experiment tracking platforms. Provides auditability and enables systematic comparison of model versions.

Lesson 2 • ML Pipelines and Workflow Orchestration

Automates data ingestion, training, and evaluation steps using pipeline orchestration tools. Reduces manual intervention and enforces reproducibility across model iterations.

Lesson 3 • Production Monitoring and Drift Detection

Detects data drift, concept drift, and performance degradation in live model deployments. Closes the MLOps loop by triggering retraining when model quality degrades.

Lesson 4 • Building REST APIs for Models

Wraps trained models in REST endpoints using lightweight web frameworks. Enables real-time inference consumption by downstream applications and services.