基于大数据分析的时空模式识别方法  

Spatiotemporal pattern recognition method based on big data analysis

作　　者：张亚姝 ZHANG Yashu(Beijing Big Data Center,Beijing 101100,China)

机构地区：[1]北京市大数据中心,北京101100

出　　处：《无线互联科技》2025年第6期16-19,共4页Wireless Internet Science and Technology

摘　　要：随着城市信息化建设深入推进,时空数据呈现大规模增长趋势,对数据处理分析技术提出更高要求。传统时空模式识别方法在处理海量数据时存在计算效率低下的问题。针对这一现状,文章提出基于分布式计算框架的多尺度时空模式识别方法,通过改进数据预处理算法提升数据质量,利用分布式存储架构解决海量数据存储难题。实验采用某大型城市3年交通轨迹数据进行验证,结果表明该方法在百万级数据处理中准确率达92.3%,计算效率提升4.3倍,为大规模时空数据分析提供新思路。With the deepening of urban informatization construction,spatiotemporal data shows a large-scale growth trend,placing higher demands on data processing and analysis technology.Traditional spatiotemporal pattern recognition methods face low computational efficiency when processing massive data.To address this situation,a multi-scale spatiotemporal pattern recognition method based on distributed computing framework is proposed,which improves data quality through enhanced data preprocessing algorithms and solves massive data storage problems using distributed storage architecture.The experiment uses three years of traffic trajectory data from a large city for verification.Results show that this method achieves 92.3% accuracy in processing million-level data with a 4.3-fold improvement in computational efficiency,providing new insights for large-scale spatiotemporal data analysis.

关 键 词：时空大数据 时空模式识别 分布式计算 多尺度分析 数据挖掘 

分 类 号：P208[天文地球—地图制图学与地理信息工程]