Intermediate: End-to-End Machine Learning

Entry requirements

This course is suitable for those who are competent in programming using Python (including notebooks, packages, data manipulation, design and use of pipelines, model evaluation functions) but are not expert programmers. The Introduction to End-to-End Machine Learning is an excellent basis for this short course.

The focus of the course is the use of released and curated Python machine learning code, rather than implementing algorithms from scratch.

Coursework involves creating code to solve a specific problem, together with a short report that describes the approach taken and critically evaluates the results. This code can be developed either using learners’ equipment (such as a laptop or PC), or with cloud-based tools such as CoLab and Kaggle Notebooks, or Jupyter notebook. A good internet connection is more important than powerful computational equipment.

The time commitment for this course is typically six to eight hours per week.

Months of entry

January, April, July, October

Course content

This short course is aimed at professionals who are seeking to understand the concepts and technologies that underpin modern machine learning.

Special offer

You can study any of our End-to-End Machine Learning short courses this July for just £950 by using the code MachineLearningJuly25 at checkout. Book before 7 July to get access to the course for this discounted price.

Course details

In this course you will learn about modern machine learning methods through five topics:

• Classification explains how best to predict discrete classes for example, accept or reject credit applications
• Training Models introduces the methods used to solve the core optimisation problem: which variant of a class of models has the least error?
• Trees and Random Forests explores how tree models can be derived, extended and deployed to produce models with validated estimates of performance on new data instances
• Dimensionality Reduction covers the rationale for and methods applicable to reducing the number of features used in predictive machine learning models
• Unsupervised Learning considers how to learn and deploy models for which there is no target variable

Advanced Python code is supplied and explained for each topic. Your key learning outcomes are to determine what models are applicable for different data and objectives, and to conduct hyperparameter-tuning or model-selection as appropriate to the model.

Who is this course for?

The course is aimed at professionals with a high level of numeracy who are seeking to understand the core concepts, methods and technologies that underpin modern machine learning.

The topics explain the key methods used to derive models that will reliably and robustly predict new and unseen instances.

The ability to contribute to such workflows is a core skill in many fields, including:

• finance (fraud prevention and credit decisions)
• healthcare (diagnostic and prognostic decisions)
• marketing (targeted ads and customer retention)

Teaching format

This is a self-paced online learning short course with lecture content, interactive elements, and access to a masterclass with the course leader after completion of the course.

Fees and funding