Prediction of Migration Outcome Using Machine Learning

The decision to migrate is mainly influenced by a combination of factors such as income, security, politics, lifestyle, education, and environmental pressures. Problems faced by Bangladeshi migrants include finding an appropriate matching job, lack of information on migration opportunities and risks; discrimination, exploitation, and abuse while overseas; and insufficient services. Traditional migration model as such gravity and radiation model based on population and distance features focused. These models are not enough for complicated migration dynamics. It is challenging to identify appropriate migration destinations for individuals since demand and supply change rapidly. We prepared a machine learning model with survey data where 1000+ individual migration worker participants from 10+ countries (the major countries where Bangladeshi workers migrate). Then applied decision tree and random forest classification to predict the successful migration outcome. The result shows that the decision tree model produces more accuracy than a random forest with this small data set. However future work should continue with large data set to explore potential machine learning techniques to find appropriate migration destination where to remain migrant satisfied and happy with the job.