Concepts for decision makers

Machine learning course: understand, brief and question machine learning, with optional code

This machine learning course explains how machine learning works, from training data to predictions and evaluation, so you can judge where it fits your business, brief a project and question the results. It is a machine learning course for beginners and managers that starts without code, with optional Code lessons in Python for training and evaluating simple models. It does not teach production deployment, and it is not a full-time bootcamp or a degree.

For you or your whole team

Start with a useful first step.

7-day free trial. No credit card needed.

You will continue on the Tosheo.ai app to create your account and review the trial terms.

What you will learn

The modules in this course.

Explaining machine learning

Describe how ML learns from examples and how it differs from generative AI.

Judging training data

Ask whether the examples a model learned from fit the decision it supports.

Understanding predictions

Tell classification, number prediction and ranking apart in business terms.

Reading evaluation results

Question accuracy claims and recognise overfitting.

Spotting bias and drift

Identify where historic data and changing conditions make a model unfair or stale.

Briefing an ML project

Define the decision, success measure and cost of errors before work starts.

Example task / fictional source material

Capstone: a decision memo on an ML proposal

You receive a fictional supplier's proposal for a model that predicts which delivery orders will arrive late. Write a memo of up to 400 words for your director recommending go, no-go or pilot, covering the training data, how performance was measured, the cost of wrong predictions, bias or drift risks and who reviews the model's output.

Review criterion

Correct concepts

Uses training data, test data and evaluation terms accurately.

Review criterion

Evaluation challenged

Questions the headline metric and asks for a measure that fits the business cost of errors.

Review criterion

Risks named

Identifies at least one specific bias or drift risk for this use case.

Review criterion

Clear recommendation

Gives a decision with conditions a director could act on.

Copy and adapt

Prompts to try

Replace the parts in square brackets with your own details, and check every output before you use it.

Translate a vendor claim

A vendor says their model has [metric and value] on [task]. Explain in plain English what that means, what it does not tell me, and five follow-up questions I should ask. Assume the base rate is [percentage, if known].

Check if ML fits

We want to [business goal]. Explain whether this needs machine learning, simple rules or a better report, and list the data we would need for each option. Keep it under 250 words and flag any assumptions.

Find hidden bias

A model will use historic [type of records] to predict [outcome]. List ways past decisions in that data could make the model unfair, and a practical check a non-technical team could request for each.

Every lesson

18 lessons you can start today.

Your diagnostic picks the order and skips what you already know. Start any lesson with a 7-day free trial, no credit card needed.

The world of machine learningExplain how machines learn from examples
FoundationStart this lesson
Training data and why it mattersExplain how data quality and bias shape a model
FoundationStart this lesson
Classification and predictionExplain how a model sorts things or predicts a number
FoundationStart this lesson
Clustering and segmentationExplain how customers or items are grouped
FoundationStart this lesson
Recommendation systemsExplain how recommendations are made and where they fail
FoundationRead the free lesson Start this lesson
Evaluating a modelRead accuracy, precision and recall without being misled
PractitionerStart this lesson
Overfitting and underfittingExplain why a model can look great and fail in use
PractitionerRead the free lesson Start this lesson
Neural networks in plain wordsExplain layers, training and why deep learning needs data
PractitionerRead the free lesson Start this lesson
Machine learning in your industryList realistic uses and limits for your sector
FoundationStart this lesson
Briefing an ML projectWrite a project brief with goal, data, success measure and risks
AdvancedStart this lesson
Questioning an ML vendorAsk the questions that expose weak claims
AdvancedStart this lesson
Fairness and bias in modelsSpot and challenge unfair outcomes
PractitionerStart this lesson
Your first model with scikit-learnTrain and test a simple classifier
PractitionerCodeStart this lesson
Regression in codeFit and evaluate a regression model
PractitionerCodeStart this lesson
Evaluating models in codeCompute accuracy, precision, recall and a confusion matrix
PractitionerCodeStart this lesson
Preparing data for MLSplit, encode and scale features correctly
PractitionerCodeStart this lesson
A neural network you can readTrain a small network and inspect its learning
AdvancedCodeStart this lesson
Using pre-trained modelsCall an existing model instead of training one
PractitionerCodeStart this lesson

Who is this machine learning course for?

It is for product managers, operations and finance leads, founders and anyone asked to approve, buy or work alongside a machine learning system. You might be reviewing a vendor's churn model, deciding whether a forecasting project is worth funding, or explaining an ML feature to customers.

Who is this machine learning course not for?

It is not for people who already work as machine learning engineers or data scientists and want advanced specialist training. The core path has no maths derivations, and the optional Code lessons in Python start from zero rather than at research level. If you want a full-time route into a machine learning job, look for a university module or an engineering-focused bootcamp; there is no job guarantee here. This course is for the people around those builders, and for anyone who wants a first hands-on start.

What will I learn, module by module?

The modules build one mental model of machine learning, then apply it to decisions you actually face.

1. What machine learning is

Learning patterns from examples instead of writing rules, and how it differs from generative AI and simple automation.

2. Training data

Labels, features and history, and why a model can only be as good as the examples it learned from.

3. Models and predictions

Classification, prediction of numbers and ranking, explained with business examples rather than equations.

4. Evaluation and overfitting

Test data, accuracy versus precision and recall in plain terms, and why a model that is perfect in testing can fail in use.

5. Bias and drift

How historic data bakes in unfairness, and why models get worse as customers and markets change.

6. Briefing and questioning an ML project

Defining the decision, the success measure and the cost of errors, and the questions to put to vendors and teams.

How does the personalised path adapt to me?

The diagnostic finds out whether you are new to the topic or have already sat through vendor demos, and what decisions you make. A product manager's examples revolve around recommendation and ranking features; a finance lead's around forecasting and fraud flags. If you can already explain training versus test data, you move straight to evaluation.

What do I practise?

You practise reading, briefing and challenging ML work. Tasks are marked against a visible rubric with specific feedback, and you can flag a score for review.

Example task: a fictional software company, Lumen Ledger, is shown a supplier's churn model that is 95 percent accurate. Only 4 percent of customers actually churn. You explain why the accuracy figure says little on its own, and list what you would ask for instead. The rubric checks that you spot that predicting 'no churn' for everyone would score 96 percent, ask how many real churners the model catches, ask what was used as test data and ask how often it will be retrained.

How long does the machine learning course take?

Each lesson is 10 to 15 minutes. Your path's length depends on your diagnostic and how regularly you learn, so there is no fixed hour total. People who already understand data and statistics often move faster through the early modules.

What do I get at the end?

You finish with a final practical assessment: reviewing an ML proposal and writing a short decision memo. Passing earns a recognised certificate in machine learning foundations for professionals. It is not an accredited qualification and does not certify engineering skills.

How much does it cost?

Machine learning foundations are included with each plan. Individual is USD 79 a month or USD 790 a year. Team is USD 189 a month or USD 1,890 a year covering up to 3 members, plus USD 49 a month or USD 490 a year for each additional member, and teams of 25 or more get volume pricing. A 7-day free trial comes first, and no credit card is needed.

Before you decide

A few good questions.

Is this a machine learning course for beginners?

Yes. It assumes no prior knowledge of machine learning, maths or coding, and builds the concepts with business examples from your role.

Will I learn to build machine learning models?

Optionally. The core path teaches how models work so you can brief and question them, without code. The optional Code lessons in Python have you train and evaluate simple models. For production-grade model building, a university module or an engineering bootcamp goes further.

How is machine learning different from ChatGPT?

ChatGPT is a generative AI tool that produces text. Many business machine learning systems instead predict or classify, such as flagging fraud or forecasting demand. The course explains both and where they overlap.

What is overfitting, in simple terms?

Overfitting is when a model learns its training examples too closely, including their quirks, so it looks excellent in testing but performs poorly on new cases.

Is the certificate accredited?

No. After a final practical assessment you earn a recognised certificate. It is not an accredited qualification or proof of engineering skill.

Can I try it free?

Yes. Every plan starts with a 7-day free trial, and no credit card is needed.

Give your team a path, not another playlist.

Explore role-specific lessons, practical tasks and the manager view.

Start free trial7-day free trial · No credit card needed

Machine learning course: understand, brief and question machine learning, with optional code

Start with a useful first step.

The modules in this course.

Explaining machine learning

Judging training data

Understanding predictions

Reading evaluation results

Spotting bias and drift

Briefing an ML project

Capstone: a decision memo on an ML proposal

Correct concepts

Evaluation challenged

Risks named

Clear recommendation

Prompts to try

Translate a vendor claim

Check if ML fits

Find hidden bias

18 lessons you can start today.

Who is this machine learning course for?

Who is this machine learning course not for?

What will I learn, module by module?

1. What machine learning is

2. Training data

3. Models and predictions

4. Evaluation and overfitting

5. Bias and drift

6. Briefing and questioning an ML project

How does the personalised path adapt to me?

What do I practise?

How long does the machine learning course take?

What do I get at the end?

How much does it cost?

A few good questions.

Keep exploring.

Guide to the AI skills professionals need

Individual and Team pricing

Every course on Tosheo.ai

Data science foundations course

Generative AI course

Give your team a path, not another playlist.