Data science fundamentals

What you will be able to do

• How the data scientist role differs from an ML engineer, and what Finnish ads ask for
• What changed in pandas 3 and NumPy 2, and how to avoid the old habits
• Splitting data first, setting a baseline and choosing a metric that fits the decision
• Telling a result in plain words, with its limits
• Using Colab and its Gemini helper without sharing personal data
• The EU AI Act points a junior needs, with the dates as of October 2026

The situation you work through

Noora, a made-up junior analyst at a made-up bike-share company in Tampere, reads four data scientist ads, then cleans a file of daily rentals with pandas 3 in Colab, splits it by time, compares a baseline with a simple scikit-learn model, and tells her manager what the model can and cannot do.

Your practice task

Pick a table with no personal data, at least 500 rows and an outcome you can predict. Load it in Colab and print the library versions. Clean it with pandas 3 habits and write a short note on what you dropped and why. Split before any other step, fit a baseline and one simple model in a pipeline, and choose the metric that matches the decision. Submit your notebook, the cleaning note, the baseline and model scores with the reason for your metric, a four-part slide (decision, number, uncertainty, risk), and one sentence each on whether the project could fall in Annex III of the EU AI Act and on personal data.

Key topics

1. The data scientist role in Finnish ads
2. Colab and the Python toolkit
3. Load and check the data with pandas 3
4. Clean the data without old habits
5. Split first, then set a baseline
6. Fit a simple model and judge it honestly

and 4 more pages

What you get when you pass

A certificate with your name, the course and the date. Sharing it is your choice. If you share it, it has a public link that anyone can open to check it, and you can add it to LinkedIn.