In this paper, we consider the prediction task for a process with nearly periodic property, i.e., patterns occur with some regularities but no exact periodicity. We propose an inference approach based on probabilistic Markov framework utilizing motif-driven transition probabilities for...