- Data, Statistics and AI for Well-Being of People and Organisations
- A cura di Luigi Fabbris, Alberto Franci, Pietro Renzi
Mitigating academic dropout through ML: The case study of the University of Florence
- Stefano Masini
- Silvia Bacci
- Alessandra Petrucci
- Bruno Bertaccini
- © 2026 Author(s) |
- CC BY 4.0
- DOI: 10.36253/979-12-215-1062-1.52
In Italy, student dropout represents one of the most critical challenges facing the higher education system. According to the ANVUR 2023 Report, the dropout rate between the first and second year of bachelor’s degree programs has shown a significant increase for students enrolled in the academic year 2020/21, rising to 14.5%, thereby returning to levels observed a decade ago. Dropout risks, however, are not limited to the first year: 20.4% of students abandon their studies within three years of enrollment in bachelor’s programs, increasing to 24.2% after six years. At the University of Florence, first-year dropout rates for bachelor’s programs exceed the national average, although a slight decline has been observed over the last four years (from 27.1% in 2020/21 to 25.3% in 2023/24). Contributing factors include insufficient academic preparation, lack of motivation, economic difficulties, and inadequate guidance, all resulting in a misalignment between student expectations and the academic experience. Dropout entails not only a substantial cost in terms of public and private resources, but also negatively affects human capital development and delays entry into the labour market. Given this context, the adoption of effective preventive policies—such as targeted pre-admission guidance and ongoing academic tutoring—is increasingly urgent. This study proposes a decision-support tool for university governance, aimed at the early identification of students at risk of dropping out. A machine learning model has been developed based on an ensemble of XGBoost, well-suited for handling complex problems with strong variable interactions, and TabNet, which provides advanced neural capabilities and selective attention mechanisms. Trained on enrollment cohorts at the University of Florence from 2020/21 to 2022/23, the model has demonstrated exceptionally high accuracy (approximately 97%) in predicting what has been observed in the 2023/24 academic year for all types of course program.
- Keywords:
- Dropout rate,
- machine learning,
- XGBoost,
- TabNet,
University of Florence, Italy - ORCID: 0000-0001-9864-5760
University of Florence, Italy - ORCID: 0000-0001-8097-3870
University of Florence, Italy - ORCID: 0000-0001-9952-0396
University of Florence, Italy - ORCID: 0000-0002-5816-2964
- AA.VV. (2023). Rapporto sul sistema della formazione superiore e della ricerca 2023. ANVUR. https://www.anvur.it/sites/default/files/2024-12/Sintesi-Rapporto-ANVUR-2023.pdf
- Almeida, L.B. (2020). Multilayer perceptrons. In E. Fiesler, R. Beale (eds.). Handbook of Neural Computation. CRC Press. pp. C1-C2. DOI: 10.1887/0750303123/b365c48
- Arik, S., Pfister, T. (2021). TabNet: Attentive interpretable tabular learning. Proceedings of the AAAI Conference on Artificial Intelligence. 35. pp. 6679-6687. DOI: 10.1609/aaai.v35i8.16826
- Caruana, R., Niculescu-Mizil, A., Crew, G., Ksikes, A. (2004). Ensemble selection from libraries of models. In Proceedings of the Twenty-First International Conference on Machine Learning (ICML '04). Association for Computing Machinery, New York, NY, USA, 18. Doi: 10.1145/1015330.1015432 DOI: 10.1145/1015330.1015432
- Chen, T., Guestrin, C. (2016). XGBoost: A scalable tree boosting system. In Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining (KDD '16). Association for Computing Machinery, New York, NY, USA, 785–794. Doi: 10.1145/2939672.2939785 DOI: 10.1145/2939672.2939785
- Dietterich, T.G. (2000). Ensemble methods in machine learning. In: Multiple Classifier Systems. MCS 2000. Lecture Notes in Computer Science, vol. 1857. Springer, Berlin, Heidelberg. Doi: 10.1007/3-540-45014-9_1 DOI: 10.1007/3-540-45014-9_1
- Ke, G., Meng, Q., Finley, T., Wang, T., Chen, W., Ma, W., Ye, Q., Liu, T. (2017). LightGBM: A highly efficient gradient boosting decision tree. In Proceedings of the 31st International Conference on Neural Information Processing Systems (NIPS'17). Red Hook (NY, USA): Curran Associates Inc., pp. 3149–3157.
- Lundberg, S.M., Lee, S. (2017). A unified approach to interpreting model predictions. In Proceedings of the 31st International Conference on Neural Information Processing Systems (NIPS'17). Red Hook (NY, USA): Curran Associates Inc., pp. 4768–4777.
- Prokhorenkova, L., Gusev, G., Vorobev, A., Dorogush, A.V., Gulin, A. (2018). CatBoost: Unbiased boosting with categorical features. In Proceedings of the 32nd International Conference on Neural Information Processing Systems (NIPS'18). Red Hook (NY, USA): Curran Associates Inc., pp. 6639– 6649.
- Wolpert, D.H. (1992). Stacked generalization. Neural Networks, 5(2), pp. 241-259. DOI: 10.1016/s0893-6080(05)80023-1
Informazioni sul capitolo
Titolo del capitolo
Mitigating academic dropout through ML: The case study of the University of Florence
Autori
Stefano Masini, Silvia Bacci, Alessandra Petrucci, Bruno Bertaccini
Lingua
Inglese
DOI
10.36253/979-12-215-1062-1.52
Opera sottoposta a peer review
Anno di pubblicazione
2026
Copyright
© 2026 Author(s)
Licenza d'uso
Licenza dei metadati
Informazioni bibliografiche
Titolo del libro
Data, Statistics and AI for Well-Being of People and Organisations
Curatori
Luigi Fabbris, Alberto Franci, Pietro Renzi
Opera sottoposta a peer review
Numero di pagine
420
Anno di pubblicazione
2026
Copyright
© 2026 Author(s)
Licenza d'uso
Licenza dei metadati
Editore
Firenze University Press
DOI
10.36253/979-12-215-1062-1
ISBN Print
979-12-215-1061-4
eISBN (pdf)
979-12-215-1062-1
eISBN (xml)
979-12-215-1063-8
Collana
Applied Statistics Series