Free Python and Data Science Courses Worth Your Time

There is no shortage of Python and data courses. The harder problem is picking a few that are good, current, and genuinely free to use rather than free for a week. This is the short list I would give a colleague in finance, operations or reporting who wants to move from spreadsheets towards Python and SQL.

Every link below was checked for availability while writing this post. Where a provider states that something is free, I say so; where there is an optional paid extra, I note it. The aim is not completeness but a sensible order of work.

Stage one: learn the language

CS50’s Introduction to Programming with Python (Harvard)

This is the one I would start most beginners on. CS50P is taught by David Malan and assumes no prior programming. The course site says anyone can take it for free through the OpenCourseWare version, working through ten weeks of material. The topics matter for data work: functions, exceptions, unit tests, regular expressions and file input and output. Testing and regex are exactly what you need when validating messy inputs. A verified certificate from edX is a separate, optional route, so ignore it unless you want the credential.

Best for: true beginners who want a clear weekly structure and problem sets to practise on.

The official Python Tutorial

The tutorial on docs.python.org is free and part of the official Python documentation, so it tracks the current release. Note its own caveat: it is aimed at programmers who are new to Python, not people new to programming altogether. If you already write VBA, SQL procedures or any scripting language, it is the fastest route to idiomatic Python.

Best for: people with some programming background who want accurate fundamentals without video.

freeCodeCamp

freeCodeCamp is a 501(c)(3) charity, and its FAQ states that the courses, projects and certifications are all free. The current curriculum lists a Python certification and a Relational Databases certification, among others. Lessons run in the browser, so there is nothing to install. One warning: its older Data Analysis with Python course is marked as no longer updated, so prefer the recommended curriculum.

Best for: learners who like building small projects in the browser before setting up a local environment.

Stage two: work with data

Kaggle Learn

Kaggle’s short courses are the most efficient way I know to get hands-on with data tables. The Pandas course lists an estimated four hours and states that it costs nothing, like all Kaggle Learn courses. Companion courses cover Python, Intro to SQL and Intro to Machine Learning. You do the exercises in notebooks hosted by Kaggle, so setup is minimal, though you will need a free account.

Best for: analysts who learn by doing and want pandas and SQL habits quickly.

The pandas getting-started tutorials

Once you have used pandas, the project’s own getting-started tutorials and 10 minutes to pandas are the reference to return to. They are short, example-driven, and match the current API.

Best for: anyone replacing Excel reconciliations and data checks with scripts.

Stage three: machine learning

Google’s Machine Learning Crash Course

Google describes Machine Learning Crash Course as a practical introduction with animated videos, interactive visualisations and exercises. Its modules include linear and logistic regression, a classification module covering thresholds, confusion matrices, precision and recall, and a module on datasets, generalisation and overfitting. Those are the concepts that matter most when your target is rare, as with fraud flags.

Best for: people who know basic Python and want the core ideas before writing models.

An Introduction to Statistical Learning

The authors offer An Introduction to Statistical Learning as a free download, with R and Python editions. It is a book rather than a course, and it is the one I would pick if you want to understand why methods work rather than just how to call them.

Best for: readers with some statistics who want depth and exercises.

fast.ai: Practical Deep Learning for Coders

The fast.ai course describes itself as free and designed for people with some coding experience. It is a deeper dive into neural networks, so I would treat it as optional unless your work is moving that way.

Best for: confident Python users interested in deep learning.

A suggested order

How to get value from any of them

• Use your own data early. After a few lessons, rebuild one small manual check from your work, such as comparing two extracts, in a script.
• Write tests for your checks. It is the habit CS50P teaches best and the one that makes automation trustworthy.
• Finish one resource before starting another. A completed short course beats five abandoned long ones.

Course websites change. If a link moves, search for the course title on the provider’s own site rather than trusting a third-party mirror.