Author(s)

Xinbing Fang, Nanping Mao, Yan Su, Hansheng Zhang, Qiang Wang, and Xiaodong Sun, Xudong Zhang, Lanhui Zeng

Corresponding Author

Xinbing Fang

Abstract

Apache Spark is an emerging engine for big data processing, developed by AMPLab at the University of California, Berkeley. Its main feature is to provide a clustered Distributed memory abstract RDD (Resilient Distributed Dataset), an immutable set of records with partitions, and a programming model in Spark. RDD in Spark has two types of operations, transformations and actions. Transformation refers to an operation that results in a new RDD, action refers to getting a numerical result through operation. Based on Spark programming framework, an online motion pattern recognition algorithm for moving objects is proposed, which takes the characteristics of stopping points into consideration. The experimental results show that the algorithm improves the recognition accuracy with the increase of training trajectory, and improves the execution speed of the algorithm in the big data environment.

Keywords

Spark; Stay points; classification