顾及缺失值的因果图时空预测网络

doi:10.11947/j.AGCS.2023.20220021

摘要/Abstract

摘要： 时空预测是地理时空大数据挖掘的基础研究命题。目前,多种模型用于预测未知系统的时空状态。然而,存在的大多数预测模型仅在没有缺失数据的时空数据集上进行测试,忽略了缺失值对预测结果的影响。在真实场景中,由于传感器或网络传输故障,数据缺失是一个不容忽视的问题。鉴于此,本文提出了一种顾及缺失值的因果图卷积网络(causal graph convolutional network considering missing values,Causal-GCNM)模型用于时空预测。Causal-GCNM模型可以自动捕捉时空数据中的缺失模式,使得Causal-GCNM模型在不需要借助额外插值算法的前提下,可以直接完成时空预测任务。本文提出的模型在3种真实的时空数据集(交通流数据集、PM_2.5监测数据集及气温监测数据集)得到了验证。试验结果表明,Causal-GCNM模型在4种缺失条件(20%随机缺失、20%块状缺失、40%随机缺失及40%块状缺失)下仍然具有较好的预测性能,并在预测精度和计算效率两类指标上优于10种存在的基线方法。

关键词: 地理时空大数据挖掘, 因果卷积网络, 图卷积网络, 时空预测, 时空数据缺失

Abstract: Spatio-temporal prediction is one of the basic research topics of geographic spatio-temporal big data mining. There are many attempts to predict spatio-temporal state of unknown systems using various deep learning algorithms. However, most existing prediction models are only tested on spatio-temporal data assuming no missing data entries, ignoring the impact of missing values on the prediction results. In the actual scenarios, data missing is an inevitable problem due to sensor or network transmission failures. Therefore, we propose a novel causal graph convolutional network considering missing values (Causal-GCNM) for spatio-temporal prediction. The proposed model can automatically capture missing patterns in the spatio-temporal data, enabling the Causal-GCNM model to directly complete the spatio-temporal prediction task without additional interpolation. The proposed model was validated on three real spatio-temporal datasets (traffic flow dataset, PM_2.5 monitoring dataset, and temperature monitoring dataset). Experimental results show that the Causal-GCNM model has good prediction performance under four missing scenarios (20% random missing, 20% block missing, 40% random missing, 40% block missing), and outperforms ten existing baseline methods in terms of prediction accuracy and computational efficiency.

Key words: geographic spatio-temporal big data mining, causal convolution network, graph convolution network, spatio-temporal prediction, spatiotemporal data missing

中图分类号:

P208

王培晓, 张彤, 聂士超, 杨瑾萱, 王天骄. 顾及缺失值的因果图时空预测网络[J]. 测绘学报, 2023, 52(5): 818-830.

WANG Peixiao, ZHANG Tong, NIE Shichao, YANG Jinxuan, WANG Tianjiao. A causal graph convolutional network considering missing values for spatio-temporal prediction[J]. Acta Geodaetica et Cartographica Sinica, 2023, 52(5): 818-830.

参考文献

[1] XU Fulong, LI Zishen, ZHANG Kefei, et al. An investigation of optimal machine learning methods for the prediction of ROTI[J]. Journal of Geodesy and Geoinformation Science, 2020, 3(2):1-15.
[2] ZHANG Xiaohong, REN Xiaodong, WU Fengbo, et al. Short-term prediction of ionospheric TEC based on ARIMA model[J]. Journal of Geodesy and Geoinformation Science, 2019, 2(1):9-16.
[3] HAN Bing, ZHAO Guoze, WANG Lifeng, et al. Earthquake electromagnetic precursor anomalies detected by a new ground-based observation network[J]. Journal of Geodesy and Geoinformation Science, 2021, 4(1):116-123.
[4] 王培晓, 张恒才, 王海波, 等. ST-CFSFDP:快速搜索密度峰值的时空聚类算法[J]. 测绘学报, 2019, 48(11):1380-1390.DOI:10.11947/j.AGCS.2019.20180538. WANG Peixiao, ZHANG Hengcai, WANG Haibo, et al. Spatial-temporal clustering by fast search and find of density peaks[J]. Acta Geodaetica et Cartographica Sinica, 2019, 48(11):1380-1390.DOI:10.11947/j.AGCS.2019.20180538.
[5] 赵琨, 罗力, 杨凤芸. 利用北斗CORS反演大气可降水量的精度分析[J]. 测绘科学, 2021, 46(11):12-17. ZHAO Kun, LUO Li, YANG Fengyun. Accuracy analysis of precipitable water vapor retrieved by BeiDou CORS[J]. Science of Surveying and Mapping, 2021, 46(11):12-17.
[6] 段功豪, 牛瑞卿, 赵艳南, 等. 基于动态指数平滑模型的降雨诱发型滑坡预测[J]. 武汉大学学报(信息科学版), 2016, 41(7):958-962. DUAN Gonghao, NIU Ruiqing, ZHAO Yannan, et al. Rainfall-induced landslide prediction based on dynamic exponential smoothing model[J]. Geomatics and Information Science of Wuhan University, 2016, 41(7):958-962.
[7] 冯宁, 郭晟楠, 宋超, 等. 面向交通流量预测的多组件时空图卷积网络[J]. 软件学报, 2019, 30(3):759-769. FENG Ning, GUO Shengnan, SONG Chao, et al. Multi-component spatial-temporal graph convolution networks for traffic flow forecasting[J]. Journal of Software, 2019, 30(3):759-769.
[8] CHENG Shifen, LU Feng, PENG Peng. Short-term traffic forecasting by mining the non-stationarity of spatiotemporal patterns[J]. IEEE Transactions on Intelligent Transportation Systems, 2021, 22(10):6365-6383.
[9] DEVRIES P M R, VIÉGAS F, WATTENBERG M, et al. Deep learning of aftershock patterns following large earthquakes[J]. Nature, 2018, 560(7720):632-634.
[10] 张显峰, 崔伟宏. 集成GIS和细胞自动机模型进行地理时空过程模拟与预测的新方法[J]. 测绘学报, 2001, 30(2):148-155. ZHANG Xianfeng, CUI Weihong. Integrating GIS with cellular automaton model to establish a new approach for spatio-temporal process simulation and prediction[J]. Acta Geodaetica et Cartographic Sinica, 2001, 30(2):148-155.
[11] 李静, 刘海砚, 郭文月, 等. 基于深度学习的人群活动流量时空预测模型[J]. 测绘学报, 2021, 50(4):522-531.DOI:10.11947/j.AGCS.2021.20200230. LI Jing, LIU Haiyan, GUO Wenyue, et al. A spatio-temporal network for human activity prediction based on deep learning[J]. Acta Geodaetica et Cartographica Sinica, 2021, 50(4):522-531.DOI:10.11947/j.AGCS.2021.20200230.
[12] 吴华意, 黄蕊, 游兰, 等. 出租车轨迹数据挖掘进展[J]. 测绘学报, 2019, 48(11):1341-1356.DOI:10.11947/j.AGCS.2019.20190210. WU Huayi, HUANG Rui, YOU Lan, et al. Recent progress in taxi trajectory data mining[J]. Acta Geodaetica et Cartographica Sinica, 2019, 48(11):1341-1356.DOI:10.11947/j.AGCS.2019.20190210.
[13] YU Bin, SONG Xiaolin, GUAN Feng, et al. K-Nearest neighbor model for multiple-time-step prediction of short-term traffic condition[J]. Journal of Transportation Engineering, 2016, 142(6):04016018.
[14] WU Shanhua, YANG Zhongzhen, ZHU Xiaocong, et al. Improved k-nn for short-term traffic forecasting using temporal and spatial information[J]. Journal of Transportation Engineering, 2014, 140(7):04014026.
[15] ZHANG J, ZHENG Y, QI D. Deep spatio-temporal residual networks for citywide crowd flows prediction[C]//Proceedings of 2017 AAAI conference on artificial intelligence.San Francisco:AAAI,2017.
[16] SHI X, GAO Z, LAUSEN L, ET AL. Deep learning for precipitation nowcasting:a benchmark and a new model[C]//Proceedings of the 31st International Conference on Neural Information Processing Systems. Guangzhou:IEEE,2017:5622-5632.
[17] KIPF T N, WELLING M. Semi-supervised classification with graph convolutional networks[C]//International Conference on Learning Representations.Toulon:ICLR, 2017.
[18] ZHAO Ling, SONG Yujiao, ZHANG Chao, et al. T-GCN:a temporal graph convolutional network for traffic prediction[J]. IEEE Transactions on Intelligent Transportation Systems, 2020, 21(9):3848-3858.
[19] YU B, YIN H, ZHU Z. Spatio-temporal graph convolutional networks:a deep learning framework for traffic forecasting[C]//Proceedings of the 27th International Joint Conference on Artificial Intelligence. Stockholm:[s.n.], 2018:3634-3640.
[20] ZHANG Yang, CHENG Tao, REN Yibin, et al. A novel residual graph convolution deep learning model for short-term network-based traffic forecasting[J]. International Journal of Geographical Information Science, 2020, 34(5):969-995.
[21] 方孟元, 唐炉亮, 杨雪, 等. 基于低频GNSS轨迹的转向级城市交通信息精细预测[J]. 测绘学报, 2021, 50(11):1469-1477.DOI:10.11947/j.AGCS.2021.20210252. FANG Mengyuan, TANG Luliang, YANG Xue, et al. Fine-grained traffic information prediction at the turning-level based on low-frequency GNSS trajectory data[J]. Acta Geodaetica et Cartographica Sinica, 2021, 50(11):1469-1477.DOI:10.11947/j.AGCS.2021.20210252.
[22] 李明晓, 张恒才, 仇培元, 等. 一种基于模糊长短期神经网络的移动对象轨迹预测算法[J]. 测绘学报, 2018, 47(12):1660-1669.DOI:10.11947/j.AGCS.2018.20170268. LI Mingxiao, ZHANG Hengcai, QIU Peiyuan, et al. Predicting future locations with deep fuzzy-LSTM network[J]. Acta Geodaetica et Cartographica Sinica, 2018, 47(12):1660-1669.DOI:10.11947/j.AGCS.2018.20170268.
[23] 樊子德, 龚健雅, 刘博, 等. 顾及时空异质性的缺失数据时空插值方法[J]. 测绘学报, 2016, 45(4):458-465.DOI:10.11947/j.AGCS.2016.20150123. FAN Zide, GONG Jianya, LIU Bo, et al. A space-time interpolation method of missing data based on spatiotemporal heterogeneity[J]. Acta Geodaetica et Cartographica Sinica, 2016, 45(4):458-465.DOI:10.11947/j.AGCS.2016.20150123.
[24] 樊子德. 异质时空数据插值方法研究[J]. 测绘学报, 2017, 46(5):668. DOI:10.11947/j.AGCS.2017.20170003. FAN Zide. Spatio-temporal interpolation methods for heterogeneous spatio-temporal data[J]. Acta Geodaetica et Cartographica Sinica, 2017, 46(5):668.DOI:10.11947/j.AGCS.2017.20170003.
[25] 程诗奋, 彭澎, 张恒才, 等. 异质稀疏分布时空数据插值、重构与预测方法探讨[J]. 武汉大学学报(信息科学版), 2020, 45(12):1919-1929. CHENG Shifen, PENG Peng, ZHANG Hengcai, et al. Review of interpolation, reconstruction and prediction methods for heterogeneous and sparsely distributed geospatial data[J]. Geomatics and Information Science of Wuhan University, 2020, 45(12):1919-1929.
[26] YANG Jinming, PENG Zhongren, LIN Lei. Real-time spatiotemporal prediction and imputation of traffic status based on LSTM and Graph Laplacian regularized matrix factorization[J]. Transportation Research Part C:Emerging Technologies, 2021, 129:103228.
[27] CHENG Shifen, LU Feng, PENG Peng, et al. Short-term traffic forecasting:an adaptive ST-KNN model that considers spatial heterogeneity[J]. Computers, Environment and Urban Systems, 2018, 71:186-198. DOI:10.1016/j.compenvurbsys.2018.05.009.
[28] CHENG Shifen, LU Feng, PENG Peng, et al. Multi-task and multi-view learning based on particle swarm optimization for short-term traffic forecasting[J]. Knowledge-Based Systems, 2019, 180:116-132.
[29] CAI Ling, JANOWICZ Krzysztof, MAI Gengchen, et al. Traffic transformer:capturing the continuity and periodicity of time series for traffic forecasting[J]. Transactions in GIS, 2020, 24(3):736-755. DOI:10.1111/tgis.12644.
[30] YI Z, LIU X, MARKOVIC N, et al. Inferencing hourly traffic volume using data-driven machine learning and graph theory[J]. Computers, Environment and Urban Systems, 2021, 85:101548. DOI:10.1016/j.compenvurbsys.2020.101548.
[31] CHEN Xinyu, SUN Lijun. Bayesian temporal factorization for multidimensional time series prediction[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2022, 44(9):4659-4673.
[32] YU H F, RAO N, DHILLON I S. Temporal regularized matrix factorization for high-dimensional time series prediction[C]//Proceedings of the 30th International Conference on Neural Information Processing Systems. Barcelona:ACM Press, 2016:847-855.
[33] CUI Zhiyong, LIN Longfei, PU Ziyuan, et al. Graph Markov network for traffic forecasting with missing data[J]. Transportation Research Part C:Emerging Technologies, 2020, 117:102671. DOI:10.1016/j.trc.2020.102671.
[34] CHE Zhengping, PURUSHOTHAM Sanjay, CHO Kyunghyun, et al. Recurrent neural networks for multivariate time series with missing values[J]. Scientific Reports, 2018, 8(1):6085. DOI:10.1038/s41598-018-24271-9.
[35] TIAN Yan, ZHANG Kaili, LI Jianyuan, et al. LSTM-Based traffic flow prediction with missing data[J]. Neurocomputing, 2018, 318:297-305. DOI:10.1016/j.neucom.2018.08.067.
[36] BAI Shaojie, KOLTER J.Zizo, KOLTUN Vladlen. An empirical evaluation of generic convolutional and recurrent networks for sequence modeling[J/OL]. arXiv:1803.01271[cs], 2018[2021-05-05]. http://arxiv.org/abs/1803.01271.
[37] YAN Jining, MU Lin, WANG Lizhe, et al. Temporal convolutional networks for the advance prediction of ENSO[J]. Scientific Reports, 2020, 10(1):8055. DOI:10.1038/s41598-020-65070-5.
[38] MA Hao, CHEN Chao, ZHU Qing, et al. An ECG signal classification method based on dilated causal convolution[J]. Computational and Mathematical Methods in Medicine, 2021, 2021:6627939.
[39] AYODEJI A, WANG Z, WANG W, et al. Causal augmented ConvNet:a temporal memory dilated convolution model for long-sequence time series prediction[J]. ISA transactions, 2022, 123:200-217.
[40] ZHENG Y, YI X, LI M, et al. Forecasting fine-grained air quality based on big data[C]//Proceedings of the 21th ACM SIGKDD international conference on knowledge discovery and data mining.Sydney:2015:[s.n.], 2267-2276.
[41] HERSBACH H, BELL B, BERRISFORD P, et al. ERA5 hourly data on single levels from 1979 to present[J]. Copernicus climate change service (c3s) climate data store (cds), 2018, 10:24381.DOI:10.24381/cds.adbb2d47.
[42] WANG Peixiao, ZHANG Tong, ZHENG Yueming, et al. A multi-view bidirectional spatiotemporal graph network for urban traffic flow imputation[J]. International Journal of Geographical Information Science, 2022, 36(6):1231-1257.
[43] CHUNG J, GULCEHRE C, CHO K, et al. Empirical evaluation of gated recurrent neural networks on sequence modeling[C]//Proceedings of 2014 Workshop on Deep Learning. Montreal:NIPS,2014.
[44] 王培晓, 王海波, 傅梦颖, 等. 室内用户语义位置预测研究[J]. 地球信息科学学报, 2018, 20(12):1689-1698. WANG Peixiao, WANG Haibo, FU Mengying, et al. Research on semantic location prediction of indoor users[J]. Journal of Geo-Information Science, 2018, 20(12):1689-1698.
[45] 沈乐君, 游志胜, 李晓峰. 自助重要性采样用于实时多目标视觉跟踪[J]. 自动化学报, 2012, 38(10):1663-1670. SHEN Lejun, YOU Zhisheng, LI Xiaofeng. Real-time visual tracking of multiple targets using bootstrap importance sampling[J]. Acta Automatica Sinica, 2012, 38(10):1663-1670.