The consideration of time factor in mobility prediction and location recommendation

Abstract

Route prediction is the missing piece in several proposed ideas for intelligent vehicles and smart cities. This field has taken great importance recently. Because of the different problems that occur on roads, the specificity of transportation and the nature of the mobility data, various models have been proposed for effective route prediction. For instance, Markov model, sequential patternsbased models, etc. As roads problems evolve, the application of these classical methods is no longer sufficient. In this thesis, we propose two novel models for route prediction, namely, PreNext and PreGraph. The first model PreNext depends on CPT (compact prediction tree) model, thus it offers all its advantages including its lossless property that allow conserving all the data in to perform prediction, its lower storage space requirement, predicting rare cases with high accuracy, etc. Our second model PreGraph is a dependency graph-based model. PreGraph represents roads as a graph which is then used to predict the next traversing road. Unlike many prediction models, the designed models are compact and easy to be constructed, and can thus provide efficient solutions for prediction. Our proposals were compared with well-known prediction models and exhibiting quite promising results on two real- world datasets.