Search
Program Calendar
Browse By Day
Browse By Room
Search Tips
Virtual Exhibit Hall
Personal Schedule
Sign In
Risk assessment tools have been criticised for leading to biased evaluations and professional negativism. Incorporating a strengths-based and protective approach could make assessments fairer and more personalised. Previous studies in this field have typically analysed the individual contribution of each risk factor, evaluating them separately with protective factors. However, these factors do not act in isolation but rather interact with each other, potentially changing the risk of social exclusion. Recent studies emphasise the role of resilience and protective factors in their interaction with risk factors.
Methodology: Comparing different machine learning techniques to predict the risk of social exclusion, this study evaluates their performance in identifying key risk and protective factors. A sample of 209 children in residential care in Southern Spain in 2021 was selected retrospectively.
Results: After applying Simple Logistic, Decision Tree, and Random Forest models, preliminary findings suggest that Random Forest outperforms other techniques. The algorithm model identifies 37 significant risk and protective factors. Risk and protective factors within the educational, social, psychological and criminological areas are the most predictive, achieving a Correct Classification Rate (CCR) of 77.9%. Our assessment tool is novel because it uses algorithms and statistical analysis to measure the risk of social exclusion experienced by unaccompanied migrant children hosted in residential care facilities.
Conclusion: Our assessment tool is novel in that it uses algorithms and statistical analysis to measure the risk of social exclusion experienced by unaccompanied migrant children in residential care. In social care, adapting services and procedures to the best predictor is crucial for resource allocation and risk management.