Supervised Machine Learning for Modelling STEM Career and Education Interest in Irish School Children

The number of unfilled jobs in Science, Technology, Engineering and Mathematics (STEM) is predicted to rise while young people’s interest in STEM careers and education is declining. Efforts to understand this decline have identified some potentially contributing factors based on statistical correlation analysis. However, these correlations can sometimes have relatively low effect-sizes. In these cases, Machine Learning (ML) techniques may provide an alternative by uncovering more complex patterns that provide stronger predictive accuracy. In this pilot study of Irish school children aged 9-13, supervised ML techniques were applied to model interest in pursuing education and careers in STEM fields. Despite the rather low coefficients from Pearson Correlation, the ML techniques were able to predict an individual’s interest in STEM careers and education with accuracies of 72.79% and 79.88% respectively. Our results suggest that ML techniques could be an important tool in understanding young people’s interest in STEM careers and education by providing models that derive more complex relationships.