This document discusses uncertainties in big data. It presents several case studies where deep learning and multi-view learning techniques have been applied to predict things like traffic incident duration, school truancy factors, and event detection in video. However, it notes that all models have uncertainties and discusses the need to better quantify objectives, propagate uncertainties, evaluate robustness, and communicate uncertainties to users. The document advocates for a multidisciplinary approach to tackling uncertainties in big data predictions and outcomes.