Machine Learning Foundations

Course outline and learning expectations

This is a tutor-led course. The outline shows what your tutor will cover; teaching materials and examinations are provided directly to enrolled students.

What you will learn

• Explain machine learning, supervised learning and unsupervised learning in plain language
• Set up and use scikit-learn in Python or Google Colab
• Prepare features and target values for a simple dataset
• Build a linear regression model for numerical prediction
• Build a logistic regression model for binary classification
• Split data into training and testing sets
• Evaluate regression and classification models using simple metrics such as mean absolute error and accuracy
• Recognise common signs of overfitting
• Build and explain a simple end-to-end prediction model using sample data

Modules

1. 1

   Machine Learning in Plain Terms

   Understand what machine learning is and distinguish supervised from unsupervised learning using simple Uganda- and Africa-relevant examples

   What machine learning isRules-based programming vs machine learningSupervised learningUnsupervised learningFeatures and targetsEveryday African use cases
2. 2

   Setting Up scikit-learn

   Prepare a beginner-friendly Python workspace in Google Colab or a local notebook and load a small dataset

   Google Colab setupInstalling and importing scikit-learnNumPy and pandas refresherLoading a CSV datasetInspecting rows and columnsBasic data checks
3. 3

   First Regression Model

   Build a linear regression model to predict maize yield from rainfall using a small sample dataset

   Regression problemsIndependent and target variablesLinearRegression workflowFitting the modelMaking predictionsInterpreting results and limitations
4. 4

   Classification Basics

   Use logistic regression to explore whether a sample SACCO loan applicant may default

   Classification problemsBinary labelsLogisticRegression workflowFitting the classifierPredicted classes and probabilitiesResponsible interpretation of loan-risk outputs
5. 5

   Training and Testing Data

   Split data correctly and understand why models must be tested on unseen examples

   Training dataTesting datatrain_test_splitRandom stateData leakage in plain languageWhy unseen data matters
6. 6

   Evaluating a Model Simply

   Measure basic model performance and explain what the scores do and do not mean

   Mean absolute error for regressionAccuracy for classificationConfusion matrix basicsCorrect and incorrect predictionsChoosing a simple metricLimits of accuracy
7. 7

   Overfitting with a Farming Analogy

   Recognise when a model memorises training data instead of learning patterns that generalise

   OverfittingUnderfittingTraining score vs test scoreFarming analogyModel complexitySimple ways to reduce overfitting
8. 8

   Capstone Prediction Model

   Build and evaluate a simple prediction model using a small sample dataset on SACCO loan default risk

   Define the prediction questionPrepare the sample dataSelect features and targetTrain logistic regressionEvaluate resultsExplain limitations, privacy and fairnessPresent findings