The Italian education system faces persistent challenges with student retention, performance, and high dropout rates. Early Warning Systems (EWS), driven by data analytics and machine learning, offer a proactive approach to identifying students at risk. This thesis aims to develop an EWS specifically tailored to Italy’s K-12 landscape, utilizing data from the Italian Ministry of Education (MIM) and INVALSI assessments to predict students’ academic outcomes, as well as whether they are likely to dropout from education. Using ten different Machine Learning algorithms, we predicted the admission status of students being in the 2nd year of High School – the last mandatory school cycle year – in 2022, by monitoring their performance in a three-year span, their reported INVALSI scores, their socioeconomic background, and their school characteristics. The same algorithms have also been employed to predict dropout for students who attended the 9th grade in 2021, looking back at their performance at the last year of Middle School, their INVALSI results and various socioeconomic indicators. Focusing on the implementation of an EWS in the Italian education, further challenges and policy implications will be explored, offering recommendations for effective integration and future research directions. Adopting this EWS framework could enable Italian schools to reduce dropout rates through timely, targeted interventions, ultimately enhancing educational attainment levels nationwide.
Il sistema educativo italiano continua a confrontarsi con sfide significative legate alle bocciature, al rendimento scolastico e agli elevati tassi di abbandono. Gli Early Warning Systems (EWS), basati sull’analisi di dati e algoritmi di machine learning, offrono un approccio proattivo per identificare gli studenti a rischio. Questa Tesi mira a sviluppare un EWS calibrato appositamente per il contesto dell’educazione K-12 in Italia, usando dati del Ministero dell’Istruzione e del Merito (MIM) e delle prove INVALSI per predire gli esiti scolastici degli studenti, e la probabilità che abbandonino la scuola. Utilizzando dieci differenti algoritmi di Machine learning, abbiamo predetto l’esito finale degli studenti che frequentavano il secondo anno di scuole superiori di secondo grando – l’ultimo anno di scuola dell’obbligo – nel 2022, monitorando le loro prestazioni in un lasso di tempo di tre anni, i loro risultati nelle prove INVALSI, il loro contesto socioeconomica e le caratteristiche delle loro scuole. Gli stessi algoritmi sono stati utilizzati anche per predire l’abbandono scolastico degli studenti che frequentavano il nono anno scolastico nel 2021, utilizzando il loro rendimento dell’ultimo anno di scuole medie, i risultati delle prove INVALSI dello stesso anno e una serie di variabili socioeconomiche. Concentrandosi sull’implementazione di EWS nell’istruzione italiana, la Tesi esplorerà ulteriori criticità e implicazioni di carattere pubblico, e offrirà raccomandazioni per integrazioni efficaci e future direzioni di ricerca. L’adozione di questo framework EWS potrebbe permettere alle scuole di ridurre i tassi di abbandono scolastico attraverso interventi mirati e tempestivi, migliorando così il livello d’istruzione su scala nazionale.
Early Warning System (EWS) in K-12 italian education
MARANI TASSINARI, GIACOMO;Mylonopoulou, Evangelia
2023/2024
Abstract
The Italian education system faces persistent challenges with student retention, performance, and high dropout rates. Early Warning Systems (EWS), driven by data analytics and machine learning, offer a proactive approach to identifying students at risk. This thesis aims to develop an EWS specifically tailored to Italy’s K-12 landscape, utilizing data from the Italian Ministry of Education (MIM) and INVALSI assessments to predict students’ academic outcomes, as well as whether they are likely to dropout from education. Using ten different Machine Learning algorithms, we predicted the admission status of students being in the 2nd year of High School – the last mandatory school cycle year – in 2022, by monitoring their performance in a three-year span, their reported INVALSI scores, their socioeconomic background, and their school characteristics. The same algorithms have also been employed to predict dropout for students who attended the 9th grade in 2021, looking back at their performance at the last year of Middle School, their INVALSI results and various socioeconomic indicators. Focusing on the implementation of an EWS in the Italian education, further challenges and policy implications will be explored, offering recommendations for effective integration and future research directions. Adopting this EWS framework could enable Italian schools to reduce dropout rates through timely, targeted interventions, ultimately enhancing educational attainment levels nationwide.File | Dimensione | Formato | |
---|---|---|---|
2024_12_MaraniTassinari_Mylonopoulou_Tesi.pdf
accessibile in internet per tutti
Descrizione: Text of the Thesis
Dimensione
9.88 MB
Formato
Adobe PDF
|
9.88 MB | Adobe PDF | Visualizza/Apri |
2024_12_MaraniTassinari_Mylonopoulou_Executive Summary.pdf
accessibile in internet per tutti
Descrizione: Executive Summary
Dimensione
933.37 kB
Formato
Adobe PDF
|
933.37 kB | Adobe PDF | Visualizza/Apri |
I documenti in POLITesi sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/10589/231502