Moving trajectory prediction model based on double layer multi-granularity knowledge discovery

An online trajectory prediction model was proposed based on multi-granularity knowledge discovery from double layers in view of the fact that the sparsity and heterogeneity in spatial distribution of real-life moving trajectory pose a challenge to model and predict moving trajectory. The operation of multi-granularity modeling and pattern mining were conducted for existing moving trajectory data on coarse/fine grained semantic layers, respectively. A hybrid prediction of online query moving trajectory can be achieved by leveraging the manipulation of matching and output complementary of online moving trajectory based on bi-layer semantic patterns. The experimental results on real data sets show that the proposed method can effectively improve the prediction accuracy and extend the range of predictable trajectory for sparse data.