Machine Learning Fundamentals

Programme overview

Machine learning is for problems where you have many examples but can't write the rule down. In this course you build two real models in Google Colab with scikit-learn. First, a rent estimator for Lagos and Abuja listings: prepare messy data, set baselines, fit linear regression and discover why the log of rent works far better, then trees and forests, overfitting and cross-validation. Then a loan default model for a microfinance bank: logistic regression, why accuracy misleads when defaults are rare, precision, recall and AUC, and a threshold chosen from what each mistake costs. You'll explain the model, test it for unfair proxies, package it as a pipeline, write a model card and plan its monitoring. Every lesson's code runs, and every answer is checked against it.

Curriculum

1. 01What Machine Learning Is25 min

   • What machine learning isWhat machine learning can and can't do, the difference between regression and classification, the workflow every project follows, and your first look at the two datasets this course uses.
2. 02Preparing Data for a Model30 min

   • Preparing data for a modelSeparate features from the target, deal with missing values and outliers without fooling yourself, turn categories into numbers, and spot leakage before it ruins a model.
3. 03Train, Test and Baselines15 min

   • Train, test and baselinesHold back data the model never sees so you can measure it honestly, choose a sensible error measure, and set a baseline that any model must beat to be worth using.
4. 04Linear Regression30 min

   • Linear regressionFit your first model, measure it against the baseline, discover why predicting the log of the target works far better for prices, and read the model's coefficients as business facts.
5. 05Trees, Forests and Overfitting30 min

   • Trees, forests and overfittingFit decision trees and random forests, watch a model memorise its training data, control overfitting, and learn why the most complex model isn't always the best one.
6. 06Cross-validation and Tuning15 min

   • Cross-validation and tuningChoose model settings without peeking at the test set, using k-fold cross-validation, and understand why a single train/test split can mislead.
7. 07Classification and Logistic Regression20 min

   • Classification and logistic regressionPredict a yes-or-no outcome with logistic regression, work with probabilities rather than labels, and see why accuracy is a dangerous measure when one outcome is rare.
8. 08Evaluating Classifiers and Choosing Thresholds30 min

   • Evaluating classifiers and choosing thresholdsRead a confusion matrix, measure precision, recall and AUC, and choose the decision threshold from what each kind of mistake costs the business.
9. 09Explaining Models and Responsible ML30 min

   • Explaining models and responsible MLExplain what drives a model's predictions with coefficients and permutation importance, check it for unfair patterns, and plan how to monitor it once it's making real decisions.
10. 10From Model to Decision30 min

    • From model to decisionPackage a model as a pipeline that takes raw applications, score new cases with reasons, save it for others to use, and document it with a model card.
11. 11Final Project25 min

    • Final project: Ladder's credit modelPlan your final project, a complete, responsible default model and lending recommendation for a microfinance bank, and start with the questions every credit model must answer.

Assessment and certificate

You are assessed with a final assessment (pass mark 60%) and a final project. Your certificate is awarded when you meet those requirements. Paying for the programme does not award it on its own.