基于多源数据的个体活动序列多步预测

doi:10.3785/j.issn.1008-973X.2025.11.011

浙江大学学报(工学版)

2025, Vol. 59

Issue (11): 2317-2325 DOI: 10.3785/j.issn.1008-973X.2025.11.011

交通工程、土木工程

基于多源数据的个体活动序列多步预测

疏阳1(

),孙轶琳1,2,*(

),梅振宇1,张逸敏2,黄毅方2

1. 浙江大学建筑工程学院，浙江杭州 310058
2. 浙江大学工程师学院，浙江杭州 310015

Multi-step prediction of individual activity sequence based on multi-source data

Yang SHU1(

),Yilin SUN1,2,*(

),Zhenyu MEI1,Yimin ZHANG2,Yifang HUANG2

1. College of Civil Engineering and Architecture, Zhejiang University, Hangzhou 310058, China
2. Polytechnic Institute, Zhejiang University, Hangzhou 310015, China

全文: PDF(2218 KB) HTML

摘要：

针对传统的以集计数据或抽样数据在个体活动序列预测研究中的局限性，提出基于多源数据的个体活动序列多步预测方法. 基于手机信令数据和兴趣点 (POI)数据，使用考虑规则和潜在狄利克雷分配（LDA）主题建模方法推断居民活动类型，构建多日活动序列. 引入nTreeClus聚类框架，识别出杭州市居民6种典型的多日活动模式，融合交通环境中的气象、日历数据，应用时间融合Transformer (TFT) 模型进行活动序列的多步预测，微平均F1分数为87%. 与传统统计模型及主流深度学习模型进行比较，分别提升了11%、6%、3%，F1分数在“娱乐”、“个人维护”活动上较序列到序列门控循环单元（S2SGRU）模型分别提升了31%、27%，显著提高了个体活动序列多步预测的准确性.

关键词： 城市交通; 交通需求预测; 时间序列预测; 活动类型挖掘; 多源数据

Abstract:

A multi-step prediction method for individual activity sequences based on multi-source data was proposed by considering the limitations of traditional research on individual activity sequence prediction based on aggregated data or sampled data. Mobile phone signaling data and point of interest (POI) data were used. Activity types of residents were inferred by using the rule-based and latent Dirichlet allocation (LDA) topic modeling methods, and multi-day activity sequences were constructed. The nTreeClus clustering framework was introduced to identify six typical multi-day activity patterns of residents in Hangzhou City. Meteorological and calendar data in the traffic environment were integrated, and the temporal fusion transformer (TFT) model was applied to conduct multi-step prediction of activity sequences. The micro-average F1 score was 87%, which increased by 11%, 6% and 3% respectively compared with traditional statistical models and mainstream deep learning models. The F1 scores for "entertainment" and "personal maintenance" activities increased by 31% and 27% respectively compared with the sequence-to-sequence gated recurrent unit (S2SGRU) model, significantly improving the accuracy of multi-step prediction of individual activity sequences.

Key words: urban transportation traffic demand forecasting time series prediction activity type mining multi-source data

收稿日期: 2024-10-22 出版日期: 2025-10-30

U 491

基金资助: 浙江省“尖兵”“领雁”研发攻关计划资助项目（2023C01240）；国家自然科学基金资助项目（52131202）.

通讯作者: 孙轶琳 E-mail: shuyang@zju.edu.cn;yilinsun@zju.edu.cn

作者简介: 疏阳（1999—），男，硕士生，从事交通大数据的研究. orcid.org/0009-0008-7053-8988. E-mail: shuyang@zju.edu.cn

	服务
	把本文推荐给朋友
	加入引用管理器
	E-mail Alert
	作者相关文章
	疏阳
	孙轶琳
	梅振宇
	张逸敏
	黄毅方

引用本文:

疏阳,孙轶琳,梅振宇,张逸敏,黄毅方. 基于多源数据的个体活动序列多步预测[J]. 浙江大学学报(工学版), 2025, 59(11): 2317-2325.

Yang SHU,Yilin SUN,Zhenyu MEI,Yimin ZHANG,Yifang HUANG. Multi-step prediction of individual activity sequence based on multi-source data. Journal of ZheJiang University (Engineering Science), 2025, 59(11): 2317-2325.

链接本文:

https://www.zjujournals.com/eng/CN/10.3785/j.issn.1008-973X.2025.11.011 或 https://www.zjujournals.com/eng/CN/Y2025/V59/I11/2317

表 1 用户基本属性的含义表

表 2 处理后手机信令数据样例

表 3 部分杭州市POI数据示例

图 1 居民活动记录与文档生成过程的对比

表 4 居民活动链的示例

图 2 居民多日活动序列的重构

图 3 平均轮廓系数

图 4 活动序列的聚类结果

图 5 多源数据下活动序列多步预测的示意图

图 6 活动多步预测的功能图

表 5 TFT与其他方法在精确率指标上的预测性能比较

表 6 TFT与其他方法在召回率指标上的预测性能比较

表 7 TFT与其他方法在F1分数上的预测性能比较

图 7 变量的相对重要性

1	ERMAGUN A, FAN Y, WOLFSON J, et al Real-time trip purpose prediction using online location-based search and discovery services[J]. Transportation Research Part C: Emerging Technologies, 2017, 77 (4): 96- 112
2	HADJIDIMITRIOU N S, CANTELMO G, ANTONIOU C Machine learning for activity pattern detection[J]. Journal of Intelligent Transportation Systems, 2023, 27 (6): 834- 848 doi: 10.1080/15472450.2022.2084336
3	LI W, ZHANG Y, CHEN Y, et al Multi-day activity pattern recognition based on semantic embeddings of activity chains[J]. Travel Behaviour and Society, 2024, 34 (1): 100682
4	ALEXANDER L, JIANG S, MURGA M, et al Origin–destination trips by purpose and time of day inferred from mobile phone data[J]. Transportation Research Part C: Emerging Technologies, 2015, 58 (9): 240- 250
5	HUANG L, LI Q, YUE Y. Activity identification from GPS trajectories using spatial temporal POIs’ attractiveness [C]//Proceedings of the 2nd ACM SIGSPATIAL International Workshop on Location based Social Networks. San Jose: ACM, 2010: 27-30.
6	CHEN C, JIAO S, ZHANG S, et al TripImputor: real-time imputing taxi trip purpose leveraging multisourced urban data[J]. IEEE Transactions on Intelligent Transportation Systems, 2018, 19 (10): 3292- 3304 doi: 10.1109/TITS.2017.2771231
7	林楠. 基于大规模手机定位数据的居民活动链挖掘方法[D]. 深圳: 中国科学院大学(中国科学院深圳先进技术研究院), 2018. LIN Nan. Mining methods for residents' activity chains based on large-scale mobile phone positioning data [D]. Shenzhen: University of Chinese Academy of Sciences (Shenzhen Institutes of Advanced Technology, Chinese Academy of Sciences), 2018.
8	ZHAO X, LI Z, ZHANG Y, et al Discover trip purposes from cellular network data with topic modeling[J]. IEEE Intelligent Transportation Systems Magazine, 2020, 14 (4): 37- 46
9	LI Z, XIONG G, WEI Z, et al Trip purposes mining from mobile signaling data[J]. IEEE Transactions on Intelligent Transportation Systems, 2021, 23 (8): 13190- 13202
10	YANG C, YE W, ZHU R, et al Resident activity pattern recognition and comparison of six Sino-American metropolises[J]. IET Intelligent Transport Systems, 2019, 13 (3): 443- 452 doi: 10.1049/iet-its.2018.5246
11	HUANG L, XIA F, CHEN H, et al Reconstructing human activities via coupling mobile phone data with location-based social networks[J]. Travel Behaviour and Society, 2023, 33 (4): 100606
12	GOULET-LANGLOIS G, KOUTSOPOULOS H N, ZHAO J Inferring patterns in the multi-week activity sequences of public transport users[J]. Transportation Research Part C: Emerging Technologies, 2016, 64 (3): 1- 16
13	ZHAI W, BAI X, PENG Z R, et al From edit distance to augmented space-time-weighted edit distance: detecting and clustering patterns of human activities in Puget Sound region[J]. Journal of Transport Geography, 2019, 78 (7): 41- 55
14	JAHANSHAHI H, BAYDOGAN M G nTreeClus: a tree-based sequence encoder for clustering categorical series[J]. Neurocomputing, 2022, 494 (7): 224- 241
15	WANG Y, ZHANG D, LIU Y, et al Enhancing transportation systems via deep learning: a survey[J]. Transportation Research Part C: Emerging Technologies, 2019, 99 (2): 144- 163
16	CONNOR J, MARTIN R, ATLAS L Recurrent neural networks and robust time series prediction[J]. IEEE Transactions on Neural Networks, 1994, 5 (2): 240- 254 doi: 10.1109/72.279188
17	CHO K, VAN M B, GULCERHRE C, et al. Learning phrase representations using RNN encoder -decoder for statistical machine translation [C]//MOSCHITTI A, PANG B, DAELEMANS W. Proceedings of the 2014 Conference on Empirical Methods in Natural Language Processing. Doha: ACL, 2014: 1724-1734.
18	HOCHREITER S, SCHMIDHUBER J Long short-term memory[J]. Neural Computation, 1997, 9 (8): 1735- 1780 doi: 10.1162/neco.1997.9.8.1735
19	翁小雄, 任杰, 覃镇林, 等基于深度注意力模型的个体出行多步预测研究[J]. 重庆交通大学学报: 自然科学版, 2022, 41 (10): 35 WENG Xiaoxiong, REN Jie, QIN Zhenlin, et al Multi-step prediction of individual travel based on deep attention model[J]. Journal of Chongqing Jiaotong University: Natural Science, 2022, 41 (10): 35
20	VASWANI A, SHAZEER N, PARMAR N, et al. Attention is all you need [C]//Proceedings of the 31st International Conference on Neural Information Processing Systems. Long Beach: Curran Associates Inc. , 2017: 6000-6010.
21	LIM B, ARIK S O, LOEFF N, et al Temporal fusion transformers for interpretable multi-horizon time series forecasting[J]. International Journal of Forecasting, 2021, 37 (4): 1748- 1764 doi: 10.1016/j.ijforecast.2021.03.012
22	高海龙, 徐一博, 刘坤, 等基于多源数据融合的高速公路路网短时交通流参数实时预测[J]. 吉林大学学报: 工学版, 2024, 54: 155- 161 GAO Hailong, XU Yibo, LIU Kun, et al Real-time prediction of short-term traffic flow parameters in expressway network based on multi-source data fusion[J]. Journal of Jilin University: Engineering and Technology Edition, 2024, 54: 155- 161
23	LIU Y, LIU Z, JIA R DeepPF: a deep learning based architecture for metro passenger flow prediction[J]. Transportation Research Part C: Emerging Technologies, 2019, 101 (4): 18- 34
24	YANG Y, XIONG C, ZHUO J, et al Detecting home and work locations from mobile phone cellular signaling data[J]. Mobile Information Systems, 2021, 2021 (1): 5546329
25	YANG C, ZHANG Y, ZHAN X, et al Fusing mobile phone and travel survey data to model urban activity dynamics[J]. Journal of Advanced Transportation, 2020, 2020 (1): 5321385
26	孙石磊, 王超, 赵元棣基于轮廓系数的参数无关空中交通轨迹聚类方法[J]. 计算机应用, 2019, 39 (11): 3293- 3297 SUN Shilei, WANG Chao, ZHAO Yuandi Parameter-free air traffic trajectory clustering method based on silhouette coefficient[J]. Computer Applications, 2019, 39 (11): 3293- 3297

[1]	王福建,张泽天,陈喜群,王殿海. 基于多通道图聚合注意力机制的共享单车借还量预测[J]. 浙江大学学报(工学版), 2025, 59(9): 1986-1995.
[2]	矫金池,孙健,倪训友. 城市网约车与巡游出租车动态博弈及碳排放效应研究[J]. 浙江大学学报(工学版), 2025, 59(7): 1373-1384.
[3]	刘瑄昀,闫莹,於志勇,黄昉菀. 基于无标度网络的类脑储备池拓扑设计[J]. 浙江大学学报(工学版), 2025, 59(7): 1385-1393.
[4]	李劲业,李永强. 融合知识图谱的时空多图卷积交通流量预测[J]. 浙江大学学报(工学版), 2024, 58(7): 1366-1376.
[5]	陈永恒,杨绥程,李世豪,寇诗雨. 基于元胞自动机的城市快速路长距离交织区运行仿真[J]. 浙江大学学报(工学版), 2024, 58(12): 2575-2585.
[6]	马泽超,刘小明,夏汗青,王伟强,王久增,申海涛. 基于图神经网络的路面病害态势预测方法[J]. 浙江大学学报(工学版), 2024, 58(12): 2596-2608.
[7]	石庆林,张凯瑞,侯亮,郭小暄,张文博,连晓振,陈永超,穆瑞. 多源数据驱动的主控式创新框架[J]. 浙江大学学报(工学版), 2024, 58(10): 2084-2095.
[8]	马壮林,毕宇明,张锐. 换乘优惠政策下公共交通乘客换乘决策行为分析[J]. 浙江大学学报(工学版), 2023, 57(12): 2513-2523.
[9]	章军辉,郭晓满,王静贤,付宗杰,陈大鹏. 基于联合概率数据融合的多目标车辆安全跟随[J]. 浙江大学学报(工学版), 2023, 57(11): 2170-2178.
[10]	奚悦,张万斌,李培楠,刘保林,胥犇,李晓军. 城市地下空间资源质量三维精细化评价[J]. 浙江大学学报(工学版), 2022, 56(4): 656-663, 710.
[11]	王飞,徐维祥. 基于LSTM神经网络改进的路阻函数模型[J]. 浙江大学学报(工学版), 2021, 55(6): 1065-1071.
[12]	王晨霖,杨洁,居文军,顾复,陈芨熙,纪杨建. 基于智能家电的短期电力负荷预测与削峰填谷优化[J]. 浙江大学学报(工学版), 2020, 54(7): 1418-1424.
[13]	汪子龙,王柱,於志文,郭斌,周兴社. 多源数据跨国人口迁移预测[J]. 浙江大学学报(工学版), 2019, 53(9): 1759-1767.

Viewed

Full text

Abstract

Cited

Shared

Discussed