Prediction of life outcomes

While the socio-demographic factors that play an important role in human lives are well understood, accurately predicting life outcomes has not been possible. In this issue, Sune Lehmann et al. introduce a machine learning approach, based on language processing techniques, that can predict different aspects of human lives. The proposed model — called ‘life2vec’ — establishes relationships between concepts, captured by an embedding space, that form the foundation for the predictions of life outcomes. The image depicts such an embedding space as it converges, where the white dots represent individuals and white lines represent how they move as the model is optimized. The shades of blue represent the density of points: the brighter the blue, the higher the density.

See Sune Lehmann et al.