Random Forest Classifier for Real-Time Operations(NPO-51337-1)

Our team is trying to improve the ground data systems that are used to analyze if data sent from space is received by JPL completely and correctly. There are several points along the journey from space to JPL where data can get lost. Our job is to first determine whether data was lost and, if so, then determine where it was lost. To accomplish this, we first shadowed human GDSA experts to learn their techniques for determining whether a pass sent from space is complete or incomplete. We programmed some of their checks in order to reduce the number of data samples we needed to hand label. Once we had a data set with complete ground truth labels, we employed several machine learning techniques, from random forests to neural networks, in order to analyze and predict the classifications.