Home
This study applies data mining techniques and multivariate statistical analysis to the 2019 Multiple Indicator Cluster Surveys (MICS) conducted by UNICEF in six Latin American and Caribbean countries, focusing on children aged 5 to 17. The research examines the main determinants of school attendance among children with and without functional difficulties. The analysis was structured in three phases: (1) descriptive analysis through SQL queries in BigQuery, (2) confirmatory analysis using correlation coefficients between age, country, and functional condition with school attendance, and (3) predictive modelling with Boosted Trees in BigQuery ML. Results indicate that 40.1% of children present at least one functional difficulty, primarily associated with learning, concentration, and behavioural adaptation. Nevertheless, school attendance was high in both groups: 97.19% for children with functional difficulties and 98.04% for those without. Correlation coefficients were low, indicating weak linear relationships. However, predictive analysis showed that age and country are the most influential factors in predicting school attendance, while functional difficulties had a minor impact. These findings provide relevant empirical evidence for the design of data-driven educational policies, especially regarding the targeting of resources for inclusive education.