Machine learning fundamentals

What you will be able to do

Skills you add to your Herizon profile when you finish: Machine Learning, Python.

• Say what machine learning is, where it beats hand-written rules, and where it does not
• Load a table, check it, and split it before you touch a model
• Train a classification model and a regression model, and read what they learned
• Test a model honestly with a baseline, a confusion matrix, precision, recall and cross-validation
• Improve a model in the right order: features first, then the model, then tuning
• Explain what neural networks are, and when a classic model or an LLM is the better tool
• Build a small project you can show in a data analyst, data scientist or ML engineer application

Who it is for

You know basic Python: variables, functions and loops. You want to know what "machine learning" means beyond the marketing. You need no earlier machine learning or statistics. Maybe you are thinking about a data job in Finland or Sweden. This course covers the everyday tools those job ads list, not a long detour through the maths.

The situation you work through

Elina, an analyst at the Finnish web shop Nordic Mug, predicts which of 2,000 past customers reorder within 90 days, from the raw table to an honestly evaluated result.

Your practice task

Pick a table where each row has a yes or no outcome and at least 500 rows. Split it, fit a DummyClassifier baseline and one real model, and report the baseline's accuracy and recall, the four counts of the confusion matrix, and precision and recall worked out by hand. Submit your notebook or script, those numbers, and two plain sentences on what the model catches and one thing it cannot tell you.

What is inside

1. What machine learning is
2. The data comes first
3. Your toolkit: pandas and scikit-learn
4. Your first model: classification
5. Predicting numbers with regression
6. Evaluating models honestly

and 5 more pages

Finish with a short quiz to earn your certificate.