Digital Examination with Project-based Submissions

Close Icon
These competencies are explicitly fostered and described in Competence View.
These competencies are fostered in this course but are not explicitly described in Competence View. Please contact the responsible person for further information. Competencies in grey are fostered in this course but are generally not the focus of Competence View, which focusses on cross-disciplinary competencies.

Each spring semester, over 500 bachelor's students in Mechanical Engineering at ETH attend the course Stochastics and Machine Learning that introduces them to probability theory, statistics, and machine learning. The course builds on prior programming experience and combines theoretical foundations with practical applications.

Assessment is divided into a digital exam after the end of the semester, two project-based submissions, and a bonus of up to 0.25 points added to the digital exam grade for weekly exercises submitted via Code Expert. The digital exam is conducted using Moodle and Code Expert. It includes conceptual questions as well as programming tasks specifically focused on stochastic methods, allowing students to demonstrate both theoretical insight and practical problem-solving skills.

The two mandatory projects can be done in teams of up to three. The projects are assessed first by an automatic review process using Kaggle. For the second project the students have to submit short reports for their group projects, which are reviewed by the teaching team. All together, the assessment supports the course’s goal of integrating statistical thinking with applied machine learning.

All Course Assessments

Overview of the Course

What is the subject context of the course?
This course is offered in the Department of Mechanical Engineering, following the «Introduction to Computer Science» and «Computer Science II» lectures. It was developed in response to the department’s desire to integrate statistical and probabilistic thinking into the curriculum, alongside machine learning. The course is structured in two interconnected parts: one focusing on stochastic methods and the other on machine learning. The aim is to bridge statistical theory with machine learning.

What should students learn and be able to do at the end of the course?
Students should be able to apply probability theory and statistical methods in the context of machine learning. Beyond theoretical understanding, they are expected to gain hands-on experience with machine learning tools and techniques—building on the lecture «Computer Science II».

Why was the specific assessment format chosen?
The assessment is split into 2 project based submissions and a digital examination. Because hands-on experience with machine learning is difficult to assess through traditional exams, students complete practical projects. These projects are implemented using Google Colab and Kaggle for project 1, and for project 2 using various computer platforms such as JupyterHub provided by ISG (Informatic support group), Google Colab, Kaggle and openreview.net. The digital exam at the end of the semester assesses both the programming component and theoretical understanding of stochastic processes, probability theory, and machine learning. It consists of Moodle-based questions and stochastic tasks in Code Expert. Additionally, students may earn up to 0.25 bonus points on the digital exam based on their performance in weekly exercises submitted via Code Expert.

How are students prepared for the assessment?
Preparation is provided through weekly exercises, detailed feedback, and supporting lectures.

Course Description

Fact Sheet

Resources

Grading and Feedback

Staff Workload (500 Candidates)

Time Staff Investment
Administration and Coordination 0.1 FTE 1 Doctoral Teaching Assistants
Project 1
- Setup of Kaggle 1 d 2 Lecturers
- Verification of automated graded Results 1 d 2 Lecturers
Project 2
- Setup of Kaggle 1 d 2 Lecturers
- Setup of openreview.net 2 d 2 Lecturers
- Creation and Distribution of Grading Guidelines 1 d 2 Lecturers
- Review of Reports 28 d 20 Doctoral and Student Teaching Assistants
- Coordination of the Review 1 d 2 Lecturers
Session Examination
- Creating Questions 1 d 3 Lecturers
1 d 13 Doctoral Teaching Assistants
- Exam 3.5 h 8 Student Teaching Assistants
3.5 h 7 Doctoral Teaching Assistants
3.5 h 3 Lecturers
- Grading 12 h 3 Lecturers

Extra Information

  • Project 1 and 2
    • One-off effort 10 d for project 1 / 15 d for project 2:
      Setup of task description and dataset, including all necessary requirements (can be used for several years)
    • Additionally 1 d for familiarization with Kaggle
    • Additionally 2 d for familiarization openreview.net platform (for project submission and review process)
  • Project 2: Review of Reports
    • Review of approximately 200 reports by 2 reviewers each:
      • 1 doctoral teaching assistants, 1 student teaching assistants for each submission
      • About 20% of reports require a third review.
  • Project 2: Coordination of the Review
    • Sending reminders to reviewers and handling individual student inquiries regarding grading.

Shared Experience

How many times has the assessment been conducted in this format?
The current assessment format has been implemented twice. In the spring semester of 2024, the course included two coding projects with automated grading of the results alongside the digital exam. In spring 2025, the format was expanded to require a short report which was manually reviewed as part of the second project.

What contributed to the success?
The course was attended by a large number of students, which made manual grading impractical for the final exam, therefore the use of digital tools (specifically the combination of Moodle and Code Expert) proved to be an ideal solution for the final exam.

What were the challenges and how were they overcome?
One major challenge was the volume of content, which needs further adjustment to better align with the scope of a 5-credit course.

The project based submissions placed a significant workload on the teaching team. Particularly each group submission in project 2, which included detailed project reports from teams of two to three students, had to be carefully reviewed and provided with feedback. Still it was important to give students a more detailed report on their work for the second project that covered further aspects beyond the purely technical performance of their code.

For the digital exams, designing effective questions was key. Open-ended questions had to be translated into multiple-choice or calculation-based formats to enable automated grading. The grading system was kept conservative, especially regarding accepted answer ranges, to ensure accuracy.

Are there any further developments planned?
A reduction in content volume is planned, while the project-based format will be retained. There is also an intention to integrate even more components into Code Expert.

What tips would you give to lecturers planning a similar assessment?
Thorough preparation is essential, especially when designing exam questions that do not require manual correction.
It is also crucial to hire a sufficient number of doctoral teaching assistants to support the evaluation of project work, which requires careful supervision by the lecturers. Office hours to support students with their project work beyond regular lectures and exercises were greatly appreciated by the students.

The extended individual feedback to the project work is key for the students' learning.
Dr. Andreas Streich

ETH Competence Framework

Subject-specific Competencies

  • Concepts and Theories (assessed)
  • Techniques and Technologies (fostered)

Method-specific Competencies

  • Analytical Competencies (assessed)
  • Decision-making (fostered)
  • Media and Digital Technologies (fostered)
  • Problem-solving (assessed)
  • Project Management (fostered)

Social Competencies

  • Communication (fostered)
  • Cooperation and Teamwork (fostered)
  • Customer Orientation (fostered)
  • Leadership and Responsibility (fostered)
  • Sensitivity to Diversity (fostered)

Personal Competencies

  • Adaptability and Flexibility (fostered)
  • Creative Thinking (fostered)
  • Critical Thinking (assessed)
  • Integrity and Work Ethics (fostered)
  • Self-awareness and Self-reflection (fostered)
  • Self-direction and Self-management (fostered)

Overview of the ETH Competence Framework

loading