In Big Data correlation becomes more important than causation. For example instead of analyzing what causes thunder you can predict thunder occurence based on lightning as both of them have very good chance of happening together within short intervals. Lightning serves as proxy to identify and predict occurence of thunder.