The empirical rule describes how points are clustered in a normally-distributed data set. It states that 95% of data fall within two standard deviations of the mean.