Mobile Edge Computing(MEC) paradigm allows mobile devices to offload complex computations to a nearby edge server with low latency. However, in mobile edge computing environment, as a mobile user moves, accessible edge servers constantly change. Moving to another edge server’s coverage area cause offloading performance degradation because the server does not have necessary data for offloading.In this paper, we propose mobility prediction models which predict mobile users’ next locations based on their past mobility data and send necessary data for offloading to edge servers close to the predicted locations. Mobility prediction models are implemented by using Markov Model, Support Vector Regression, and Recurrent Neural Network. We evaluated mobility prediction models by performing simulations, which showed that the models can improve computation offloading in mobile edge computing environment.