# Machine Learning Foundations for Professionals

> Understand how machine learning models are built, tested and deployed, and work hands-on with real datasets in Python.

- **Provider:** Afriskora Training Solutions
- **Category:** Data, AI & Digital Skills
- **Duration:** 5 days
- **Formats:** classroom (cities across Africa), live virtual (Zoom/Teams), in-house for teams
- **Certificate:** Afriskora Certificate of Completion, verifiable online
- **Booking:** request dates and a quote; invoice and EFT payment
- **Web page:** https://afriskora.co.za/courses/data-ai-digital/machine-learning-foundations-for-professionals
- **Request dates:** https://afriskora.co.za/request-training?course=machine-learning-foundations-for-professionals

## Overview

Many organisations buy AI tools without anyone inside who understands how the models behind them learn, fail or drift. This course takes analysts, engineers and technical managers from first principles to a working model: preparing data, choosing an algorithm, measuring accuracy honestly and explaining results to decision-makers. Exercises use public African datasets from agriculture, health and finance, and no advanced mathematics is assumed.

## Who should attend

- Data and business analysts moving into machine learning
- Engineers and IT professionals supporting AI projects
- Technical managers who oversee data science teams
- Researchers and statisticians modernising their toolkit

## Learning outcomes

- Explain supervised, unsupervised and generative learning in practical terms
- Prepare and clean data so that models learn the right patterns
- Train, tune and compare models in Python with scikit-learn
- Measure model performance and spot overfitting, bias and drift
- Plan how a model moves from notebook to production safely

## Course outline

### Day 1: Machine learning in context

- Where machine learning fits alongside rules and statistics
- The model life cycle from question to deployment
- Python, notebooks and core libraries
- Framing a business problem as a learning task

### Day 2: Data preparation

- Exploring and profiling a dataset
- Handling missing values, outliers and imbalance
- Feature engineering and encoding
- Train, validation and test splits

### Day 3: Supervised learning

- Regression for forecasting quantities
- Classification with trees, ensembles and logistic models
- Hyperparameter tuning and cross-validation
- Choosing the right metric for the decision

### Day 4: Unsupervised and generative methods

- Clustering customers, assets and transactions
- Anomaly detection for fraud and faults
- How large language models and embeddings work
- When to use a pre-trained model instead of building one

### Day 5: Responsible deployment

- Explainability and communicating results
- Fairness, bias testing and data protection
- Monitoring for drift after go-live
- Capstone: present a model to a mock steering committee