【摘要】 近年来有关POI的数据挖掘是研究热点之一,但很少有专门针对POI类型间变化规律的研究工作。基于此,提出一种新的停留点提取方法用于从原始GPS轨迹中提取出用户的POI点,并基于所有用户的POI点得到一个POI类型转移矩阵序列,用基于向量自回归模型的POI类型转移演化算法BVAREA来学习类型转移矩阵的变化规律。并在Geolife数据集上进行实验,结果表明,BVAREA模型在MAE和RMSE指标上都优于基于AR、MA和SES模型,在MAE指标上预测性能最高可以分别提升9.23%、14.84%和14.1%。更多还原

【Abstract】 Data mining on POI(Point Of Interest) has been one of the research interests in recent years, but there is little research on the transfer rule of the type between types that are used to represent the semantic POI. Based on this, proposes a new method to extract the stay point, which is used to obtain the POI type sequences of the user from the original GPS data. Meanwhile, obtains the sequences about POI type transfer matrix from the POI type sequences, and BVAREA, a POI type transfer evolution algorithm based on vector autoregressive model, is used to learn the transfer rule of type transfer matrix. The experiments are conducted on the Geolife data set, the results show that the BVAREA model is superior to contrast models that based on the AR, MA and SES in terms of MAE and RMSE standards. The highest predict performance in the MAE can be improved by 9.23%, 14.84% and 14.1%, respectively.更多还原