Applied ai machine learning course

Lesson 1 • AI Landscape and Key Terminology

Map the AI ecosystem from narrow AI to general AI and position ML within it. Establishes shared vocabulary used throughout the entire course.

Lesson 2 • Mathematics Essentials for ML

Review linear algebra, calculus, probability, and statistics concepts that underpin ML algorithms. Focuses on intuition and applied use rather than formal proofs.

Lesson 3 • Python and Data Science Toolkit

Set up a reproducible Python environment and gain fluency with core data science libraries. Directly enables hands-on coding in all subsequent chapters.

Lesson 1 • Exploratory Data Analysis Techniques

Apply statistical summaries and visualisations to uncover distributions, correlations, and anomalies. EDA findings directly inform feature engineering decisions.

Lesson 2 • Data Splitting and Validation Strategy

Design train, validation, and test splits that yield unbiased performance estimates. Establishes evaluation discipline required for every modeling chapter ahead.

Lesson 3 • Handling Missing and Noisy Data

Detect, quantify, and remediate missing values, outliers, and inconsistencies. Clean data is the prerequisite for reliable model training covered later.

Lesson 1 • Encoding Categorical Variables

Convert nominal and ordinal categories into numeric representations suitable for ML algorithms. Choice of encoding directly affects model accuracy and interpretability.

Lesson 2 • Feature Selection Methods

Remove irrelevant and redundant features using filter, wrapper, and embedded techniques. Reduces overfitting and training cost for models built in later chapters.

Lesson 3 • Scaling and Normalisation

Apply standardisation, min-max, and robust scaling to numeric features. Scaling is critical for distance-based and gradient-based algorithms introduced next.

Lesson 4 • Creating and Transforming Features

Engineer interaction terms, polynomial features, and domain-derived variables to boost signal. Demonstrates how human insight amplifies algorithmic learning.

Lesson 1 • Linear and Logistic Regression

Derive and implement linear regression for continuous targets and logistic regression for binary classification. These interpretable baselines anchor all algorithm comparisons.

Lesson 2 • Decision Trees and Rule-Based Models

Build decision trees using information gain and Gini impurity criteria. Trees introduce the bias-variance trade-off central to all subsequent ensemble methods.

Lesson 3 • Evaluation Metrics and Model Selection

Choose appropriate metrics for regression and classification tasks and compare models rigorously. Correct metric selection prevents misleading conclusions in applied projects.

Lesson 1 • Bagging and Random Forests

Reduce variance by training multiple trees on bootstrap samples and aggregating predictions. Random forests are the practical starting point for ensemble modeling.

Lesson 2 • Gradient Boosting Fundamentals

Build additive models by fitting successive learners to residual errors. Gradient boosting consistently outperforms single models on structured data.

Lesson 3 • Advanced Boosting Frameworks

Apply XGBoost, LightGBM, and CatBoost to large tabular datasets with speed and accuracy. These frameworks dominate competitive ML benchmarks on structured data.

Lesson 1 • Association Rule Mining

Extract co-occurrence patterns from transactional data using Apriori and FP-Growth. Supports recommendation systems and market basket analysis use cases.

Lesson 2 • Clustering Algorithms

Partition data into meaningful groups using K-Means, DBSCAN, and hierarchical methods. Clustering drives customer segmentation and exploratory pattern discovery.

Lesson 3 • Dimensionality Reduction Techniques

Compress high-dimensional data while preserving variance or local structure using PCA and manifold methods. Enables visualisation and speeds up downstream model training.

Lesson 1 • Recurrent Networks and Sequence Modeling

Model sequential dependencies with RNNs, LSTMs, and GRUs for time-series and text data. Sequence models bridge classical ML and modern transformer architectures.

Lesson 2 • Regularisation and Generalisation

Prevent overfitting in deep networks using dropout, batch normalisation, and weight decay. Generalisation techniques are mandatory for production-grade models.

Lesson 3 • Convolutional Neural Networks

Apply convolution, pooling, and feature map hierarchies to image classification tasks. CNNs are the backbone of computer vision applications in industry.

Lesson 4 • Perceptrons and Feedforward Networks

Build intuition for neurons, layers, and activation functions as universal function approximators. Provides the architectural foundation for all deep learning models.

Lesson 1 • Model Packaging and Serving

Serialise models and expose them via REST APIs and batch inference pipelines. Serving patterns determine latency, throughput, and scalability in production.

Lesson 2 • CI/CD Pipelines for ML

Automate testing, validation, and deployment of ML models through continuous integration pipelines. CI/CD reduces manual errors and accelerates safe model updates.

Lesson 3 • Responsible AI and Governance

Embed fairness, transparency, and accountability into the ML lifecycle from design to retirement. Governance frameworks protect organisations and end users from AI-related harms.

Lesson 4 • Production Monitoring and Drift Detection

Track data drift, concept drift, and model performance degradation in live systems. Proactive monitoring prevents silent model failures that erode business outcomes.