Why?
Machine learning turns data into predictions and is the centre of modern data science. This course gives you the core algorithms and the habits needed to evaluate them honestly.
What?
A practical machine learning course: an overview of the field, logistic regression, evaluation of models, decision trees and random forests, support vector machines, k-nearest neighbours, regression, unsupervised learning (clustering and dimensionality reduction) and artificial neural networks. Taught with Python and scikit-learn labs and a final project.
Curriculum:
Machine Learning Overview
What machine learning is, the KDD process, supervised vs unsupervised learning and typical applications.
Logistic Regression
Binary classification with the sigmoid function, the cost function and gradient-based training.
Evaluation of Machine Learning Models
Train/test splits, cross-validation, confusion matrix, accuracy, precision, recall, F1 and ROC/AUC, plus overfitting and underfitting.
Decision Trees
How trees split on impurity measures (entropy, Gini), pruning and interpretation.
Random Forest
Bagging and ensembles of trees, feature importance and tuning.
Support Vector Machines
Maximum margin classification, soft margins and kernels such as RBF.
K-Nearest Neighbors
Distance based classification, choosing k and the effect of scaling.
Regression
Simple, multiple and polynomial regression and regression metrics.
Unsupervised Machine Learning
K-means and hierarchical clustering, and dimensionality reduction with PCA.
Artificial Neural Networks
Perceptron, feed forward networks, activation functions and backpropagation.
Notes
Python and pandas knowledge helps a lot. Labs use Jupyter notebooks with real datasets (heart disease, articles, bank deposits). The course project is a full classification workflow.