Assessment¶
The split is the same for all four codes and fixed by the syllabi.
| Component | Points | What it is |
|---|---|---|
| Activity in lectures and seminars | 10 | minitests in INSIS during lectures and seminars |
| Term project | 25 | 20 for the project itself, 5 for the opponent review of another team's project |
| Presentation and defence | 5 | 9 or 16 December (winter run); mandatory for every team member — each student answers a verification question on the project that checks they did the work themselves |
| Mid-term test | 15 | 11 November (before the first ÚTIA seminar), seminars 1–4; written in the computer lab, about 30 minutes; includes programming tasks based on the nbgrader assignments; internet access and AI assistants are switched off |
| Final test | 30 | a quiz covering the four seminars and the lectures; date in INSIS |
| Home assignments | 15 | online courses (DataCamp or IBM Training / SkillsBuild): 10 points; the four nbgrader assignments: 5 points in total |
Home assignments¶
-
Online courses — 10 points: choose one of the two platforms:
- DataCamp: earn at least 10,000 XP and submit DataCamp certificates issued during this course.
- IBM Training / SkillsBuild: complete at least 10 hours of courses on the IBM SkillsBuild / IBM Training platform during the semester and submit the certificates or completion records.
In both cases, certificates issued before the course do not count. * Nbgrader — 5 points: the scores from all four assignments are scaled to a maximum of 5 points in total.
Recommended DataCamp courses¶
Students may select any DataCamp courses (or, on the IBM platform, any courses in Python, data science or AI) according to their prior knowledge, academic interests and learning objectives, subject to the XP and certificate requirements above. The following courses provide a suggested starting point for students developing their foundations in Python, data analysis and machine learning. This list is advisory, not a prescribed curriculum. Students with relevant prior experience are encouraged to choose more advanced courses that extend their existing competencies; they are not required to complete introductory courses.
| Course | Main focus |
|---|---|
| Introduction to Python | Python fundamentals for students with little or no programming experience |
| Data Manipulation with pandas | Preparing, transforming and summarising tabular data |
| Introduction to Data Visualization with Matplotlib | Creating and customising data visualisations |
| Exploratory Data Analysis in Python | Exploring datasets and interpreting patterns |
| Supervised Learning with scikit-learn | Building and evaluating supervised learning models |
| Linear Classifiers in Python | Further study of linear classification methods |
| Hyperparameter Tuning in Python | Further study of model tuning and hyperparameter optimisation |
Students new to the subject should begin with Python fundamentals and data analysis before progressing to machine learning, taking account of each course's stated prerequisites.
Grading scale¶
| Grade | Points |
|---|---|
| 1 — excellent | 90 and more |
| 2 — very good | 75 – 89 |
| 3 — good | 60 – 74 |
| 4 — fail | below 60 |
The final test can be retaken according to the faculty's ECTS rules. Bonus points may be offered for additional assignments announced in the seminars.
Use of AI tools¶
Proposed policy — to be confirmed before the first assignment is due
AI tools may be used for the term project, under the faculty's rules for theses: the team is responsible for the output and must understand it, the report contains a paragraph describing the team's own contribution, and each member states which part of the work they wrote. Because AI can be used at home, the points that count are earned where it cannot: the minitests, the mid-term test, the final test and the individual defence of the project. AI assistants are switched off during all tests.
As part of the course, students receive free tokens for IBM Bob, IBM's AI coding assistant, which they may use for the term project and the home assignments under the policy above. Access details are given out in the seminars.
What the tests cover¶
- Mid-term (11 November): seminars 1–4 and their assignments — NumPy and pandas semantics (views, broadcasting, chained assignment, cleaning), leakage-free pipelines, metrics and thresholds, model selection, structured model output, throughput, embeddings, retrieval-augmented generation. Part of the test are programming tasks modelled on the nbgrader assignments — if you solved the assignments yourself, you are prepared; the material on jupyter.vse.cz is available during the test, internet and AI assistants are not.
- Final test: all four seminars, the ÚTIA deep-learning block as covered in the lectures, and the lecture material.