84 Views | January 19, 2021
Italian University system is characterized by an high level of student dispersion compared to the European standard. In particular statistical data indicate a level of failure to complete the cycle of studies which for Italy is around 40% of students while in Western European countries is between 21% and 28%. The phenomenon of student dropout has strong negative repercussions both for the university system and for the whole country in terms of non-return on investments made for the growth of skills.
The “Osservatorio Abbandoni” project was undertaken by Cineca with the aim of applying ML algorithms to produce a predictive analysis of university dropouts. The Observatory starts from data in University DataWarehouses to extract students socio-demographic, career and performance information and use them to carry out a predictive analysis on the probability of abandonment and display the results in dashboards created through Microstrategy Dossier. Thanks to the MicroStrategy Platform it was possible to create a data model integrating ML tools with historical data. In other words, it was possible to analyze in a single Dossier data coming from the historical data warehouse and predictive analysis results. The results are processed by an R engine that provides predictive data on abandonment based on different variables selected by analyzing the historical behavior of students. The recipients of the project are typically Statistical Offices, Department Directors and the University Quality Presidia. These users will use the Dossiers to analyze both historical dropout data and forecast data. The dashboards that have been created allow to analyze a particular metric (e.g. the dropout rate) according to the main dimensions of the student (age group, sex, residence, etc.) and to correlate it with particular features of his or her academic career and university (school of origin, grades obtained, progression speed, etc.).
"Throughout this, Microstrategy Dossier has helped us to create clear and easy-to-use dashboards for the end user, for example through interactive data filtering capabilities."
~Federico Gallerani, Solution Consultant at CINECA)
The possibility of knowing in advance which students may be at risk can allow the University to identify the main causes that lead to abandonment and to intervene in time to limit and / or control the phenomenon through contrast actions (direct contact, courses recovery, tutoring, questionnaires).