KNIME logo
Contact usDownload
Read time: 3 min

UNIMB & KNIME Announce Winners of ML Challenge

April 14, 2023
Teaching with KNIMECompany news
Stacked TrianglesPanel BG

Today, April 14 2023, we are pleased to announce the winner teams of the “Machine Learning Challenge” organized by the Università Milano-Bicocca and KNIME for the students of the academic year 2022/2023 enrolled in the Machine Learning course taught by Prof. Fabio Stella for the Master in Data Science from the Department of Informatics, Systems and Communication.

The competition’s theme was “Diabetes Prediction Using Machine Learning Algorithms”. More than 80 students, forming 23 teams, submitted their project solutions on February 17. KNIME’s open source data science software, KNIME Analytics Platform, was selected as the tool of choice for the challenge, for its versatility, coverage, and accessibility.

Congratulations to the Winners!

1st place was awarded to Team 18 – Luca Porcelli and Vittorio Haardt. This project was the overall winner because of the originality of their approach, centered around the end user rather than just the model performance.

1st place winners: Vittorio Haardt and Luca Porcelli

2nd place was secured by Team 11 – Stella Cervini, Daniel Montalbano, and Giuseppe Sabino. The accuracy of all steps of the approach, from data exploration, data preparation, model training, model testing, and model deployment were the highlights of this project.

2nd place winners: Daniel Montalbano, Giuseppe Sabino, and Stella Cervini,

3rd place went to Team 19 – Davide Della Libera, Carlotta Bellomo, and Marianna Primi. In this project we appreciated the effort in the application of a relatively large neural network and the nice separation between the data app for deployment and the data app for data exploration.

3rd place winners: Davide Della Libera, Carlotta Bellomo, and Marianna Primi

The judging committee was composed of two members from KNIME, and Prof. Fabio Stella from the Università Milano-Bicocca. All projects were evaluated in terms of:

  • Performance of the machine learning model, quantified via the logloss value achieved on the test set

  • Overall approach and explanation, which includes technical merit, clarity of expression and communication of ideas, appropriate referencing and context of the present work, overall balance and structure of report, repetition, diagrams, tables, captions

  • Elegant and correct implementation , that is: complexity, cleanness, and usability of the KNIME workflows

After evaluating all 23 projects the judging committee produced a shortlist of the six best projects.

The six shortlisted teams were provided with feedback to improve their projects, including better visualization of the data exploration data app, addressing data leakage issues, revising the deployment strategy, improving the model execution performance, and more. The six teams were then given two weeks to incorporate the feedback into their work and prepare the project presentation.

The last part of the challenge was the final presentation of the project details and achievements in front of the judging committee. Based on how much of the feedback had been addressed and how well the project details and achievements have been communicated, the judging committee finally decided for the top three winners of this challenge.

The winners were announced on April 13, in the Aula Magna of the Università Milano-Bicocca during a KNIME Data Connect event. The project presentations and the award ceremony have been recorded and the video is now available on YouTube.

We would like to thank all participants for the work, the ideas, the implementation, and finally for incorporating the feedback into the original projects.

It was a pleasure to work with all of you!

About the University of Milano-Bicocca: The University of Milano-Bicocca was founded in 1998 and in nearly 25 years has developed an extensive international network in various fields of research, including many world-famous universities, research centers and top corporations. It has a multidisciplinary approach that trains professionals in subjects areas such as economics, law, sciences, technology, medicine, sociology, statistics, psychology and pedagogy. The University of Milano-Bicocca has a unique campus structure and has invested over the years in advanced teaching support services, state-of-the-art laboratories, modern infrastructures.

About KNIME: KNIME software bridges the worlds of dashboards and advanced analytics through an intuitive interface, appropriate for anybody working with data. It empowers more business experts to be self-sufficient and more data experts to push the business to the bleeding edge of modern data science, integrating the latest AI and machine learning techniques. KNIME is distinct in its open approach, which ensures easy adoption and future-proof access to new technologies.

Data Science @ Scale in Higher Education

Data Science @ Scale in Higher Education

December 5, 2022 | by Tobias Schlüter
Educators Name KNIME's Top Features for Teaching

Educators Name KNIME's Top Features for Teaching

January 9, 2023 | by Rosaria Silipo
Teaching Low Code Data Science: A Lecturer’s View

Teaching Low Code Data Science: A Lecturer’s View

September 15, 2022 | by Dursun Delen