Supervised and Unsupervised Learning Techniques Training

About This Course

Machine learning has become a cornerstone of modern data-driven decision-making, and organisations across every sector are actively seeking professionals who can harness its power responsibly and effectively. Yet there remains a significant skills gap between those who understand machine learning conceptually and those who can design, train, evaluate, and interpret models in a production-ready context. TrainingComplex's Supervised and Unsupervised Learning Techniques Training programme bridges that gap with a structured, technically rigorous, and practically focused curriculum.

Over ten intensive days, participants explore the full landscape of supervised learning — including regression, classification, decision trees, ensemble methods, and support vector machines — before advancing into unsupervised paradigms such as clustering, dimensionality reduction, and anomaly detection. Each technique is introduced with the mathematical intuition necessary to understand what the algorithm is doing, followed immediately by hands-on coding exercises and real-world case studies that cement understanding.

The programme also addresses the critical but often overlooked stages of machine learning workflows: feature engineering, hyperparameter tuning, cross-validation, model comparison, and communicating results to non-technical stakeholders. Participants leave equipped not only to build models but to choose the right approach for a given problem, diagnose failure modes, and present findings in a way that drives organisational action.

Whether you are a data analyst seeking to formalise your machine learning skills, an engineer moving into an ML-focused role, or a researcher applying predictive modelling to domain-specific challenges, this programme provides the depth, structure, and practical exposure needed to perform at a professional level.

Who Should Attend

Data analysts seeking to formalise and expand their machine learning capabilities
Software engineers and developers transitioning into machine learning or data science roles
Business intelligence professionals looking to incorporate predictive modelling into their work
Researchers applying statistical and ML techniques to domain-specific datasets
IT professionals tasked with evaluating or implementing ML-based solutions
Quantitative analysts in finance, health, agriculture, or public sector organisations
Academics and postgraduate students requiring applied ML skills for research projects

Learning Objectives

Explain the theoretical foundations and practical differences between supervised and unsupervised learning
Implement and tune key supervised learning algorithms including linear regression, logistic regression, decision trees, random forests, gradient boosting, and support vector machines
Apply unsupervised techniques including K-means, hierarchical clustering, DBSCAN, PCA, t-SNE, and autoencoders to real datasets
Design robust model evaluation pipelines using cross-validation, confusion matrices, ROC curves, silhouette scores, and other appropriate metrics
Perform systematic feature engineering, selection, and transformation to improve model performance
Diagnose and address common pitfalls including overfitting, underfitting, class imbalance, and data leakage
Select the most appropriate algorithm for a given problem type, dataset size, and organisational constraint
Optimise models through hyperparameter search strategies including grid search and Bayesian optimisation
Interpret and explain model outputs using techniques such as SHAP values, partial dependence plots, and cluster profiling
Build end-to-end ML workflows using Python, scikit-learn, and supporting libraries
Communicate model findings and limitations clearly to both technical peers and non-technical decision-makers
Apply ethical considerations around bias, fairness, and transparency in machine learning

Course Modules

Module 1: Foundations of Machine Learning

The machine learning landscape: supervised, unsupervised, and semi-supervised paradigms
Key terminology: features, labels, training, validation, and test sets
The bias-variance tradeoff and its implications for model design
Data types and their relevance to algorithm selection
Setting up the Python ML environment: Anaconda, Jupyter, scikit-learn, and supporting libraries
End-to-end ML workflow overview: from problem framing to model deployment

Module 2: Data Preparation and Feature Engineering

Exploratory data analysis: distributions, correlations, and outlier identification
Handling missing values: imputation strategies and their statistical implications
Feature scaling: standardisation, normalisation, and when each applies
Encoding categorical variables: one-hot encoding, target encoding, and ordinal encoding
Feature selection methods: filter, wrapper, and embedded approaches
Detecting and preventing data leakage in preprocessing pipelines

Module 3: Linear and Logistic Regression

Simple and multiple linear regression: assumptions, estimation, and interpretation
Evaluating regression models: MAE, MSE, RMSE, and R-squared
Regularisation techniques: Ridge, Lasso, and Elastic Net
Logistic regression: the sigmoid function, log-loss, and decision boundaries
Multiclass classification with logistic regression: one-vs-rest and softmax
Diagnosing and addressing violations of regression assumptions

Module 4: Decision Trees and Ensemble Methods

Decision tree construction: splitting criteria, Gini impurity, and information gain
Controlling tree complexity: max depth, min samples, and pruning strategies
Bagging and the random forest algorithm: diversity through randomness
Gradient boosting machines: sequential error correction and learning rates
XGBoost and LightGBM: practical implementation and tuning strategies
Feature importance extraction and interpretation from tree-based models

Module 5: Support Vector Machines and Kernel Methods

The maximum-margin classifier: geometric intuition and the support vector concept
Soft-margin SVMs: the role of the C parameter in controlling misclassification tolerance
The kernel trick: transforming non-linearly separable data into higher-dimensional space
Common kernels: linear, polynomial, and radial basis function (RBF)
SVM for multi-class classification and regression (SVR)
Practical tuning and computational considerations for SVMs on large datasets

Module 6: Model Evaluation and Validation Strategies

Holdout validation, k-fold cross-validation, and stratified k-fold for imbalanced data
Classification metrics in depth: precision, recall, F1-score, and the confusion matrix
ROC curves, AUC-ROC, and precision-recall curves: choosing the right diagnostic
Handling class imbalance: oversampling (SMOTE), undersampling, and class weights
Hyperparameter optimisation: grid search, randomised search, and Bayesian optimisation
Comparing models statistically: avoiding misleading conclusions from single-run results

Module 7: Clustering Algorithms and Applications

The unsupervised learning problem: what we seek when there are no labels
K-means clustering: the algorithm, initialisation strategies, and the elbow method
Hierarchical clustering: agglomerative linkage methods and dendrogram interpretation
Density-based clustering with DBSCAN: handling noise and arbitrarily shaped clusters
Evaluating cluster quality: silhouette score, Davies-Bouldin index, and domain validation
Applied clustering: customer segmentation, document grouping, and anomaly flagging

Module 8: Dimensionality Reduction Techniques

The curse of dimensionality and its impact on distance-based algorithms
Principal Component Analysis (PCA): eigenvectors, explained variance, and scree plots
Applying PCA as a preprocessing step for supervised and unsupervised pipelines
t-SNE: non-linear dimensionality reduction for visualisation of high-dimensional data
UMAP: faster, scalable non-linear reduction with preserved global structure
Selecting the right number of components and interpreting reduced representations

Module 9: Anomaly Detection and Density Estimation

Anomaly detection use cases: fraud, fault detection, intrusion detection, and quality control
Statistical approaches: z-score, IQR-based methods, and Gaussian distribution thresholds
Isolation Forest: the logic of isolating anomalies through random partitioning
Local Outlier Factor (LOF): density-based anomaly scoring
One-class SVM for novelty detection in high-dimensional feature spaces
Evaluating anomaly detection models in the absence of fully labelled ground truth

Module 10: Introduction to Neural Networks and Deep Learning Concepts

The perceptron and the feedforward neural network: architecture and forward pass
Activation functions: ReLU, sigmoid, tanh, and their effect on gradient flow
Backpropagation and stochastic gradient descent: training neural networks in practice
Regularisation in neural networks: dropout, batch normalisation, and early stopping
Autoencoders as unsupervised representation learners and anomaly detectors
When to use neural networks versus classical ML: capacity, data volume, and interpretability trade-offs

Module 11: Model Interpretability and Explainability

The interpretability-performance trade-off in model selection
Intrinsically interpretable models: linear models, decision trees, and rule-based systems
Post-hoc global explanations: feature importance, partial dependence plots, and ALE plots
SHAP (SHapley Additive exPlanations): theory and application to black-box models
LIME: locally faithful approximations for individual prediction explanation
Communicating model explanations to non-technical stakeholders and regulators

Module 12: Ethics, Fairness, and Responsible Machine Learning

Sources of bias in data, labelling, and model design
Fairness definitions and their mathematical tensions: demographic parity, equalised odds, and calibration
Audit frameworks for detecting discriminatory model behaviour
Transparency and accountability in automated decision-making systems
Privacy considerations: differential privacy, data minimisation, and federated learning concepts
Developing and applying an organisational checklist for responsible ML deployment

Module 13: End-to-End Project and Deployment Readiness

Problem framing and stakeholder alignment: translating a business question into an ML task
Building reproducible ML pipelines with scikit-learn Pipeline and ColumnTransformer
Versioning data, code, and models for collaborative and auditable workflows
Model deployment fundamentals: REST APIs, containerisation concepts, and serving predictions
Monitoring deployed models: data drift, concept drift, and retraining triggers
Presenting findings to decision-makers: structuring the narrative, visualising results, and articulating limitations

Requirements & Prerequisites

Participants should have a working knowledge of Python programming, including familiarity with libraries such as NumPy and pandas. A foundational understanding of statistics — including concepts such as probability, distributions, and regression — is strongly recommended. Prior exposure to data analysis or a quantitative discipline is helpful, though no prior machine learning experience is required.

Training Methodology

Guided coding labs in which participants implement each algorithm from scratch and then using scikit-learn, reinforcing both intuition and practical skill
Dataset-based case studies drawn from sectors including healthcare, finance, retail, and public administration, applied throughout the programme
Algorithm comparison workshops where participants run multiple models on the same dataset and defend their choice of approach through structured peer discussion
Hyperparameter tuning challenges with leaderboard-style feedback to simulate real model optimisation pressure
Model interpretability exercises using SHAP values and visualisation tools to practise communicating results to simulated non-technical audiences
Diagnostic debugging sessions in which participants are given broken or poorly performing pipelines and must identify and fix the root cause
End-to-end mini-projects completed in small teams, covering problem framing through to model presentation
Daily reflection prompts linking technical content to participants' own organisational data challenges and use cases

Organisational Benefits

Organisations gain staff capable of designing and deploying machine learning solutions that directly support strategic objectives
Improved quality of data-driven decisions through rigorous model validation and evaluation practices
Reduced reliance on external consultants as in-house ML capability is strengthened
Faster turnaround on analytical projects as participants apply structured, repeatable ML workflows
Better risk management through participants' ability to identify and mitigate model bias and data quality issues
Enhanced cross-functional collaboration as participants can translate ML outputs into actionable business insights
Increased competitiveness through the adoption of predictive and pattern-recognition capabilities across business units
A shared internal language and methodology for machine learning projects, improving team cohesion and knowledge transfer

Where Is Supervised and Unsupervised Learning Techniques Training Offered?

This course runs in-person across: Nairobi, Kigali, Zanzibar, Abuja, Riyadh, Nakuru, Cape Town, Mombasa, Lagos, Dubai, Naivasha, Kisumu, Accra, Addis Ababa, Arusha, Pretoria, Kampala, Johannesburg, Dar es Salaam, plus Live Virtual (online) cohorts. Browse live dates for your city below.

Upcoming Schedules

Location	Format	Dates	Price
Live Virtual	Online	19 Oct 2026 – 30 Oct 2026	KES 160,000
Live Virtual	Online	24 Oct 2026 – 13 Dec 2026	KES 160,000
Nairobi, Kenya	In-Person	26 Oct 2026 – 4 Nov 2026	KES 230,000
Kigali, Rwanda	In-Person	26 Oct 2026 – 4 Nov 2026	USD 3,800
Zanzibar, Tanzania	In-Person	26 Oct 2026 – 4 Nov 2026	USD 5,000
Abuja, Nigeria	In-Person	26 Oct 2026 – 4 Nov 2026	USD 5,600
Riyadh, Saudi Arabia	In-Person	26 Oct 2026 – 4 Nov 2026	USD 9,000
Nakuru, Kenya	In-Person	26 Oct 2026 – 4 Nov 2026	KES 240,000
Cape Town, South Africa	In-Person	26 Oct 2026 – 4 Nov 2026	USD 7,600
Mombasa, Kenya	In-Person	26 Oct 2026 – 4 Nov 2026	KES 240,000
Lagos, Nigeria	In-Person	26 Oct 2026 – 4 Nov 2026	USD 5,000
Dubai, United Arab Emirates	In-Person	26 Oct 2026 – 4 Nov 2026	USD 8,200
Dubai, United Arab Emirates	In-Person	2 Nov 2026 – 13 Nov 2026	USD 8,200
Abuja, Nigeria	In-Person	2 Nov 2026 – 13 Nov 2026	USD 5,600
Kigali, Rwanda	In-Person	2 Nov 2026 – 13 Nov 2026	USD 3,800

+228 more dates — view live schedule →

Frequently Asked Questions

Do I need prior machine learning experience to attend this course?

No prior machine learning experience is required. However, participants should be comfortable programming in Python and have a basic understanding of statistics such as probability and regression. The programme is designed for intermediate-level practitioners ready to formalise their ML skills.

What software and tools will we use during the course?

The course uses Python as its primary programming language, with hands-on work in scikit-learn, NumPy, pandas, Matplotlib, and Seaborn. Participants will also be introduced to XGBoost, LightGBM, and SHAP. A laptop with Anaconda or a compatible Python environment installed is required.

Will I receive a certificate upon completing the course?

Yes. Participants who attend the full programme and complete the end-to-end project receive a Certificate of Completion from TrainingComplex, which is suitable for professional development portfolios and continuing education records.

Is this course theoretical or hands-on?

The programme is strongly hands-on. Approximately 60% of contact time is spent in guided coding labs, team mini-projects, and applied exercises using real datasets. Theory is introduced to build intuition but is always immediately reinforced through practical implementation.

Can my organisation arrange a private cohort of this training?

Yes. TrainingComplex offers customised in-house delivery of this programme for teams of six or more. The curriculum can be tailored to your organisation's specific datasets, industry context, and learning priorities. Please contact us directly to discuss scheduling and pricing.

View Live Schedule & Book →