Understand

Intermediate · after NumPy and basic statistics

Applied Machine Learning

Beta This course is a beta version. Lessons, exercises and translations are still being reviewed and may change. Report anything that looks wrong with the feedback button.

Calling .fit() takes one line. Knowing whether the result means anything takes this course. You implement the core algorithms in NumPy so they stop being magic, then use scikit-learn properly: pipelines without leakage, validation that matches deployment, metrics that match costs, and explanations whose limits you can state.

Modules
13
Lessons
30
Time
~33 hours
Projects
2
Get started

What you'll be able to do

  • Frame a prediction problem, choose a metric from the cost of errors, and set a baseline
  • Implement and explain linear and logistic regression, trees and k-means
  • Validate correctly, including time-based splits, and prevent leakage with pipelines
  • Choose thresholds, check calibration, and handle class imbalance
  • Interpret models, and state the limits of explanations and fairness metrics
  • Deliver a reproducible predictive model with a justified method

Who it's for

People who can use NumPy and pandas and know what a confidence interval is, such as graduates of Data Science & Analytics.

What you need

Python and NumPy; means, variance and basic probability. The linear algebra and calculus are taught where they are used, and every chapter is open. scikit-learn downloads tens of MB the first time an exercise uses it.

What you'll build

  • Bike Demand Forecast
  • Predictive Modelling Capstone

Curriculum

FoundationsThe mathematics and the framing. 4 modules · 10 lessons
  1. 1 Thinking in ArraysNumPy, shapes, vectorisation and the linear algebra machine learning actually uses.
  2. 2 Calculus and LikelihoodGradients you can check, and where loss functions come from.
  3. 3 Framing and BaselinesMost ML failures start here.
  4. 4 Learning from DataModels, losses, gradient descent, and evaluation that doesn't lie.
ModelsRegression, classification, trees, ensembles and neighbours. 3 modules · 8 lessons
  1. 5 Regression, ProperlyClosed form, regularisation and assumptions.
  2. 6 Logistic Regression, Thresholds and CalibrationA probability first, a decision second.
  3. 7 Trees, Ensembles and NeighboursNon-linear models, from splits to boosting.
The CraftEvaluation, features, unsupervised learning and interpretation. 4 modules · 9 lessons
  1. 8 Evaluation in DepthCross-validation, time, imbalance, and ROC versus PR.
  2. 9 Features and PipelinesEncoding, imputation and selection, inside the fold.
  3. 10 Unsupervised Learningk-means, PCA and anomalies, with no label to check you.
  4. 11 Interpretation and FairnessWhat explanations can say, and who the errors fall on.
Professional PracticeTuning, tracking, documentation, capstone. 2 modules · 3 lessons
  1. 12 Professional ML WorkflowTuning without leaks, tracking, and model cards.
  2. 13 CapstoneAn authentic predictive problem, from framing to model card.