Leveraging Aridhia TRE for Identifying Pupils at Risk of Being Not in Education, Employment or Training (NEET)

October 2, 2023 | Aygul

Project Overview

In the realm of education, identifying and supporting vulnerable pupils is a critical undertaking. One particular challenge is to predict which students are at risk of becoming NEET (Not in Education, Employment, or Training). The implications of becoming NEET at a young age are far-reaching, often leading to depression, unemployment, lower wages, and decreased quality of life. To address this issue, in summer 2023, the University of Warwick’s Data Science for Social Good program (DSSGx UK) collaborated with the EY Foundation and several local councils to develop a machine learning model and a complementary dashboard aimed at predicting pupils at risk of becoming NEET. This initiative was conducted using data from the Bradford Metropolitan District Council, focusing on potentially vulnerable young individuals.

The Aridhia Trusted Research Environment (TRE)

The Aridhia TRE played a pivotal role in facilitating the success of this machine learning project. The TRE served as a secure, efficient, and collaborative platform that enabled the researchers to achieve their goals.

Global Team Collaboration

One of the standout features of the Aridhia TRE was its capacity to foster global team collaboration. The project involved researchers from various geographical locations, including the University of Warwick, the EY Foundation, and local councils. The TRE provided a centralized hub where team members could securely collaborate, share data, and work on the project in real-time. Features such as secure data sharing and version control ensured that all stakeholders were up-to-date, regardless of their physical location.

VMs Running ML Models

A crucial component of this project was the development of machine learning models to predict which pupils were at risk of becoming NEET. The Aridhia TRE offered a robust and scalable computing environment for running these models. Virtual Machines (VMs) within the TRE allowed researchers to create and deploy their ML algorithms efficiently. The TRE’s high-performance computing capabilities ensured that complex ML models could be trained and fine-tuned without the need for additional infrastructure investments. This streamlined the development process, allowing researchers to focus on model accuracy and efficiency.

Data Security and Compliance

Given that the project involved sensitive data from potentially vulnerable young individuals, data security was crucial. The Aridhia TRE, being ISO-certified, provided the necessary safeguards to manage data risks effectively. The TRE enforced strict data access controls, ensuring that only authorized personnel could access and manipulate the data.

Outcome and Impact

The Aridhia TRE played a crucial role in the success of the project aimed at predicting NEET and supporting vulnerable pupils. By leveraging the capabilities of the Aridhia TRE, the researchers were able to develop a predictive model and accompanying dashboard that local schools can now use to provide targeted support to at-risk pupils. Furthermore, the project not only achieved its goal but also ensured the ethical and secure handling of sensitive data, ultimately benefiting the well-being and future prospects of young individuals at risk of becoming NEET.

Dashboard and Model

The dashboard solution developed for this project is a browser-based application and can be hosted in the secure Aridhia TRE. It is interactive in its nature meaning users can explore the data, for example by filtering and selecting subgroups. Key features are automatically presented in clear and concise visualisations and tables. For example, student- and school- level plots where all relevant information including attendance, attainment results, special educational needs (SENs) and other characteristics are tracked through time.

Through automating the data processing and analytics pipeline, time needed for the identification of potential NEET students is reduced compared to the time it takes to manually curate the information. The backend workings of the application have been developed in a way to allow for additional features to be easily incorporated. The project has been developed in a modular and reproducible manner which means that the visualisations we have created can be used for other projects. This means we can do similar analyses for different schools and councils.



Dr. Aygul Zagidullina joined DSSGx UK in 2023 as a Technical Mentor and was involved in the development of dashboarding and machine learning tools to help with the identification of NEET students within the UK councils. Aygul has a background in Econometrics and Statistics and previously worked in academia and industry. Her PhD was specialised in dimensionality reduction and covariance estimation techniques with broad applications in different domains. Currently, Aygul is a lecturer at the University of Applied Sciences in Luzern.

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.