A carregar...
Publicação
Beyond Intentions: A Data-Driven Model of Brazilian Migration Destinations for Decision-Making: Predicting individual migration trends to support strategic decisions
| Nome: | Descrição: | Tamanho: | Formato: | |
|---|---|---|---|---|
| 2.55 MB | Adobe PDF |
Autores
Resumo(s)
This thesis explores the motivations and challenges faced by Brazilian emigrants, aiming to
bridge the gap between descriptive migration studies and predictive modelling. While existing
research has predominantly focused on macro-level flows or qualitative analyses, this study
adopts a novel approach by building and analysing an original dataset of 173 survey responses
from Brazilian migrants to Australia and Ireland. To uncover the patterns influencing
destination choices and migration experiences — and to move beyond explanation toward
prediction — this study applies a supervised machine learning model (Random Forest),
preceded by Principal Component Analysis (PCA) for dimensionality reduction and feature
selection. Despite limitations such as a modest sample size, class imbalance, and the presence
of noisy self-reported data, the model achieves an overall accuracy of 73% in predicting
destination country, correctly identifying 81% of respondents who migrated to Australia and
60% to Ireland. The most influential features are predominantly qualitative, including
perceived cost of living, use of social media and messaging apps to access information, and
motivations such as language acquisition and job opportunities. These results suggest that
behavioural traits and perceptions outweigh sociodemographic characteristics in determining
migration outcomes. The findings offer technical insights for institutions such as universities,
immigration agencies, and private service providers targeting migrants. The model enables
the identification of undecided or influenceable migrant profiles, supports the development
of communication strategies tailored to digital information-seeking behaviours, and highlights
optimal moments for outreach. Additionally, by introducing a reproducible pipeline for
predictive modelling based on primary behavioural data, the study addresses a key
methodological gap in current migration research. By shifting the analytical lens from
aggregate flows to individual-level decisions, it demonstrates that predictive analytics can
generate meaningful insights even within small and complex datasets. This approach
contributes a new perspective to migration research and offers institutions a scalable tool to
plan strategically and respond proactively to emerging migration trends based on data, rather
than assumptions.
Descrição
Dissertation presented as the partial requirement for obtaining a Master's degree in Data Driven Marketing, specialization in Marketing Intelligence
Palavras-chave
Brazilian Migration Predictive Modelling PCA Random Forest Migration Prediction SDG 4 - Quality education SDG 8 - Decent work and economic growth SDG 10 - Reduced inequalities