懸浮微粒(統稱PM，含有粗及細懸浮微粒)對健康的影響不亞於其他任何污染物。其中細懸浮微粒(PM2.5)因粒徑小可深入肺泡，並可能抵達細支氣管壁，干擾肺部內的氣體交換。長期暴露於懸浮微粒，可引發心血管病、呼吸道疾病以及增加肺癌的危險。本研究期望以資料探勘基礎, 提供適當和足夠的解釋力，可有效的探討細懸浮微粒資料之時空異質性及趨勢性，及展示各個地區細懸浮微粒的特性與空間分佈，喚起公眾對環境健康之重視。本研究組合多種類神經包含卷積類神經網路與長短期記憶等，並考量測站時序資料包含前六個時刻PM2.5、PM10、溫度、風速、風向、平均風速、平均風向與相對溼度，以及高程空間資訊。在模型設計上，透過多個測站時序整體趨勢的表現，且考慮相鄰測站與相似測站的關聯性。在本研究中，我們提出了兩種選擇相關測站的方式，相鄰測站採用kNN-ED找出空間關聯性，而相似測站採用kNN-DTWD找出時序關聯性。透過類神經網路將兩者關聯性合併且透過資料訓練其關聯比例，進而可解釋該測站之表現

Particulate matter (PM) has a greater impact on human health than any other contaminants. The small diameter of fine particulate matter (PM2.5) allows it to penetrate deep into the alveoli as far as the bronchioles, thereby interfering with the exchange of gas within the lungs. Long-term exposure to particulate matter has been shown to cause cardiovascular disease, respiratory disease, and increase the risk of lung cancer. In this study, we employed data mining to reveal the characteristics of PM2.5 and its spatial distribution in various regions of the world. We also aim to make forecasts of air quality for durations of up to 48 hours using a combination of multiple neural networks including a Convolutional Neural Network and Long-Short Term Memory. The proposed predictive model takes into account a variety of meteorology data from the previous six hours, including PM2.5, PM10, temperature, wind speed, wind direction, average wind speed, average wind direction, and relative humidity. We even consider information related to the elevation space. The design of the model design accounts for overall trends at multiple stations, the correlation between adjacent stations, and the correlations among stations. kNN-ED is used to derive spatial correlations between adjacent stations, whereas kNN-DTWD is used to derive correlations between time series at various stations. The proposed neural network considers both kNN-ED and kNN-DTWD correlations, with the weighting specific to each station determined using training data. The proposed model was evaluated using two datasets from Taiwan and Beijing. The scheme outlined in this paper has recently been implemented by the Taiwan Environmental Protection Administration to provide hourly PM2.5 forecasts for all cities in Taiwan.

[1] Evaluation and application of the short-range(0-6hr) PQPFs from an ensemble prediction system based on LAPS, http://etd.lib.nctu.edu.tw/cgibin/gs32/ncugsweb.cgi/login?o=dncucdr&s=id=”GC976401005”.&searchmode=basic.
[2] Government open data : Air quality stations info., http://data.gov.tw/node/6075.
[3] QQ Air Quality(QQAQ), http://qqaq.ee.ncku.edu.tw.
[4] TWEPA Instruments, https://taqm.epa.gov.tw/taqm/tw/b0102-3.aspx.
[5] 財團法人氣象應用推廣基金會, http://www.metapp.org.tw/index.php/weatherknowledge/37-typhoon/83-2009-01-22-08-04-48.
[6] R. Agrawal and S. Ieong. Aggregating web offers to determine product prices. In Proceedings of the 18th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, pages 435–443, 2012.
[7] R. Agrawal and R. Srikant. Fast algorithms for mining association rules in large databases. In Proceedings of the 20th International Conference on Very Large Data Base, pages 487–499, 1994.
[8] R. Agrawal and R. Srikant. Mining sequential patterns. In Proceedings of the 12nd IEEE International Conference on Data Engineering, pages 3–14, 1995.
[9] S. Aseervatham, A. Osmani, and E. Viennet. bitSPADE: A lattice-based sequential pattern mining algorithm using bitmap representation. In Proceedings of the 16th IEEE International Conference on Data Mining, pages 792–797, 2006.
[10] J. Ayres, J. Flannick, J. Gehrke, and T. Yiu. Sequential pattern mining using a bitmap representation. In Proceedings of the 18th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, pages 429–435, 2002.
[11] E. Bakshy, I. Rosenn, C. Marlow, and L. Adamic. The role of social networks in information diffusion. In Proceedings of the 21st International Conference on World Wide Web, pages 519–528, 2012.
[12] W. R. Barnard. Evaluating the contribution of pm2.5 precursor gases and re-entrained road emissions to mobile source pm2.5 particulate matter emissions. 2004.
[13] B. S. Beckerman, M. Jerrett, M. Serre, R. V. Martin, S.-J. Lee, A. Van Donkelaar, Z. Ross, J. Su, and R. T. Burnett. A hybrid approach to estimating national scale spatiotemporal variability of pm2.5 in the contiguous united states. Environmental science & technology, 47(13):7233–7241, 2013.
[14] H.-P. Cao, N. Mamoulis, and D.-W. Cheung. Mining frequent spatio-temporal sequential patterns. In Proceedings of the 15th IEEE International Conference on Data Mining, pages 82–89, 2005.
[15] H.-W. Chen, C.-T. Tsai, C.-W. She, Y.-C. Lin, and C.-F. Chiang. Exploring the background features of acidic and basic air pollutants around an industrial complex using data mining approach. Chemosphere, 81(10):1358–1367, 2010.
[16] L. J. Chen, Y. H. Ho, H. H. Hsieh, S. T. Huang, H. C. Lee, and S. Mahajan. Adf: an anomaly detection framework for large-scale pm2.5 sensing systems. IEEE Internet of
Things Journal, PP(99):1–1, 2017.
[17] Y.-C. Chen, S.-H. Chang, W.-C. Peng, and S.-Y. Lee. Efficient algorithms for influence maximization in social networks. Knowledge and Information Systems, 33(3):577–601, 2012.
[18] Y.-C. Chen, C.-C. Chen, W.-C. Peng, and W.-C. Lee. Mining correlation patterns among appliances in smart home environment. In Proceedings of the 18th Pacific-Asia Conference on Knowledge Discovery and Data Mining, pages 222–233, 2014.
[19] Y.-C. Chen, W.-C. Peng, and S.-Y. Lee. Exploring community structures for influence maximization in social networks. In Proceedings of the 6th International Workshop on Social Network Mining and Analysis, joint with the 18th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, pages 1–6, 2012.
[20] Y.-C. Chen, W.-C. Peng, and W.-C. Lee. A novel system for mining useful correlation in smart home. In Proceedings of the 6th International Workshop on Domain Driven Data Mining, joint with the 23th IEEE International Conference on Data Mining, pages 357–364, 2013.
[21] Y.-C. Chen, W.-Y. Zhu, W.-C. Peng, W.-C. Lee, and S.-Y. Lee. CIM: Community-based
influence maximization in social networks. ACM Transactions on Intelligent Systems and Technology, 5(2):1–31, 2014.
[22] H. Cheng, X.-Y. Yan, and J.-W. Han. IncSpan: Incremental mining of sequential patterns in large database. In Proceedings of the 10th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, pages 527–532, 2004.
[23] Y.-S. Choi, C.-H. Ho, D. Chen, Y.-H. Noh, and C.-K. Song. Spectral analysis of weekly
variation in pm10 mass concentration and meteorological conditions over china. Atmospheric Environment, 42(4):655–666, 2008.
[24] H.-J. Chu, C.-Y. Lin, C.-J. Liau, and Y.-M. Kuo. Identifying controlling factors of ground-level ozone levels over southwestern taiwan using a decision tree. Atmospheric environment, 60:142–152, 2012.
[25] H.-J. Chu, H.-L. Yu, and Y.-M. Kuo. Identifying spatial mixture distributions of pm2.5 and pm10 in taiwan during and after a dust storm. Atmospheric environment, 54:728–737, 2012.
[26] W.-H. Du, J.-W. Rau, J.-W. Huang, and Y.-S. Chen. Improving the quality of tags using state transition on progressive image search and recommendation system. In Proceedings of the IEEE International Conference on Systems, Man, and Cybernetics, pages 3233–3238, 2012.
[27] L. Ferrero, G. Mocnik, B. Ferrini, M. Perrone, G. Sangiorgi, and E. Bolzacchini. Vertical profiles of aerosol absorption coefficient from micro-aethalometer data and mie calculation over milan. Science of The Total Environment, 409(14):2824 – 2837, 2011.
[28] E. Fix and J. L. Hodges Jr. Discriminatory analysis-nonparametric discrimination: consistency properties. Technical report, DTIC Document, 1951.
[29] M.-N. Garofalakis, R. Rastogi, and K. Shim. SPIRIT: Sequential pattern mining with
regular expression constraints. In Proceedings of the 25th International Conference on Very Large Data Base, pages 223–234, 1999.
[30] K. Greff, R. K. Srivastava, J. Koutn´ık, B. R. Steunebrink, and J. Schmidhuber. LSTM: A search space odyssey. CoRR, abs/1503.04069, 2015.
[31] P. Haider, L. Chiarandini, and U. Brefeld. Discriminative clustering for market segmentation. In Proceedings of the 18th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, pages 417–425, 2012.
[32] J.-W. Han, J. Pei, and Y.-W. Yin. Mining frequent patterns without candidate generation. ACM Special Interest Group on Management of Data, 29(2):1–12, 2000.
[33] K. He, X. Zhang, S. Ren, and J. Sun. Delving deep into rectifiers: Surpassing human-level performance on imagenet classification. In Proceedings of the 2015 IEEE International Conference on Computer Vision (ICCV), ICCV ’15, pages 1026–1034, Washington, DC, USA, 2015. IEEE Computer Society.
[34] C.-C. Ho, H.-F. Li, F.-F. Kuo, and S.-Y. Lee. Incremental mining of sequential patterns over a stream sliding window. In Proceedings of the International Workshop on Mining Evolving and Streaming Data, joint with 16th IEEE International Conference on Data Mining, pages 677–681, 2006.
[35] S. Hochreiter and J. Schmidhuber. Long short-term memory. Neural Comput., 9(8):1735–1780, Nov. 1997.
[36] C.-S. Horng, C.-A. Huh, K.-H. Chen, P.-R. Huang, K.-H. Hsiung, and H.-L. Lin. Air
pollution history elucidated from anthropogenic spherules and their magnetic signatures in marine sediments offshore of southwestern taiwan. Journal of Marine Systems, 76(4):468–478, 2009.
[37] Y.-H. Hu, C.-K. Huang, H.-R. Yang, and Y.-L. Chen. On mining multi-time-interval
sequential patterns. Data and Knowledge Engineering, 68(10):1112–1127, 2009.
[38] J.-W. Huang, S.-C. Lin, and M.-S. Chen. DPSP: Distributed progressive sequential pattern mining on the cloud. In Proceedings of the 16th Pacific-Asia Conference on Knowledge Discovery and Data Mining, pages 27–34, 2010.
[39] J.-W. Huang, C.-Y. Tseng, M.-C. Chen, and M.-S. Chen. PISAR: Progressive image search and recommendation system by auto-interpretation and user behavior. In Proceedings of the IEEE International Conference on Systems, Man, and Cybernetics, pages 1442–1447, 2011.
[40] J.-W. Huang, C.-Y. Tseng, J.-C. Ou, and M.-S. Chen. A general model for sequential
pattern mining with a progressive database. IEEE Transactions on Knowledge and Data Engineering, 20(9):1153–1167, 2008.
[41] Y. Hwa-Lung and W. Chih-Hsin. Retrospective prediction of intraurban spatiotemporal distribution of pm2. 5 in taipei. Atmospheric Environment, 44(25):3053–3065, 2010.
[42] A. Kurt and A. B. Oktay. Forecasting air pollutant indicator levels with geographic models 3days in advance using neural networks. Expert Systems with Applications, 37(12):7986–7992, 2010.
[43] Y. Lecun, L. Bottou, Y. Bengio, and P. Haffner. Gradient-based learning applied to document recognition. In Proceedings of the IEEE, pages 2278–2324, 1998.
[44] V. C.-C. Liao and M.-S. Chen. DFSP: a depth-first spelling algorithm for sequential pattern mining of biological sequences. Knowledge and Information Systems, 38(3):623–639, 2014.
[45] C.-Y. Lin, S. C. Liu, C. C.-K. Chou, S.-J. Huang, C.-M. Liu, C.-H. Kuo, and C.-Y. Young. Long-range transport of aerosols and their impact on the air quality of taiwan. Atmospheric Environment, 39(33):6066–6076, 2005.
[46] C.-Y. Lin, Z. Wang, W.-N. Chen, S.-Y. Chang, C. C. Chou, N. Sugimoto, and X. Zhao. Long-range transport of asian dust and air pollutants to taiwan: observed evidence and model simulation. Atmospheric Chemistry and Physics, 7(2):423–434, 2007.
[47] J. Lin, E. Keogh, L.Wei, and S. Lonardi. Experiencing sax: a novel symbolic representation of time series. Data Mining and Knowledge Discovery, 15(2):107–144, Oct 2007.
[48] M.-Y. Lin and S.-Y. Lee. Incremental update on sequential patterns in large databases by implicit merging and efficient counting. Information System, 29(5):385–404, 2004.
[49] C.-M. Liu, C.-Y. Young, and Y.-C. Lee. Influence of asian dust storms on air quality in taiwan. Science of the Total Environment, 368(2):884–897, 2006.
[50] X.-M. Liu, J. Biagioni, J. Eriksson, Y. Wang, G. Forman, and Y.-M. Zhu. Mining largescale, sparse GPS traces for map inference: Comparison of approaches. In Proceedings of the 18th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, pages 669–677, 2012.
[51] S.-J. Lu, D. Wang, X.-B. Li, Z. Wang, Y. Gao, and Z.-R. Peng. Three-dimensional distribution of fine particulate matter concentrations and synchronous meteorological data measured by an unmanned aerial vehicle (uav) in yangtze river delta, china. Atmospheric Measurement Techniques Discussions, 2016:1–19, 2016.
[52] Y. Mao, W.-L. Chen, Y.-X. Chen, C.-Y. Lu, M. Kollef, and T. Bailey. An integrated data mining approach to real-time clinical monitoring and deterioration warning. In Proceedings of the 18th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, pages 1140–1148, 2012.
[53] F. Masseglia, P. Poncelet, and M. Teisseire. Incremental mining of sequential patterns in large databases. Data and Knowledge Engineering, 46(1):97–121, 2003.
[54] N. Moustafa, G. Creech, E. Sitnikova, and M. Keshk. Collaborative anomaly detection framework for handling big data of cloud computing. CoRR, abs/1711.02829, 2017.
[55] S. Nguyen and M. Orlowska. Improvements of INCSPAN: Incremental mining of sequential patterns in large database. In Proceedings of the 9th Pacific-Asia Conference on Knowledge Discovery and Data Mining, pages 442–451, 2005.
[56] J. S. Park, M.-S. Chen, and P. S. Yu. An effective hash-based algorithm for mining
association rules. SIGMOD Rec., 24(2):175–186, May 1995.
[57] S. Parthasarathy, M. Zaki, M. Ogihara, and S. Dwarkadas. Incremental and interactive sequence mining. Proceedings of the 18th ACM International Confereince on Information and Knowledge Management, 1999.
[58] J. Pei, J.-W. Han, B. Mortazavi-asl, H. Pinto, Q. Chen, U. Dayal, and M.-C. Hsu. PrefixSpan: Mining sequential patterns efficiently by prefix-projected pattern growth. In Proceedings of the 17th IEEE International Conference on Data Engineering, pages 215–224,2001.
[59] S. Qin, F. Liu, C. Wang, Y. Song, and J. Qu. Spatial-temporal analysis and projection of extreme particulate matter (pm10 and pm2.5) levels using association rules: A case study of the jing-jin-ji region, china. Atmospheric Environment, 120(Supplement C):339 – 350,2015.
[60] I. Shafer, K. Ren, V. Boddeti, Y. Abe, G.-R. Ganger, and C. Faloutsos. RainMon: An integrated approach to mining bursty timeseries monitoring data. In Proceedings of the 18th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, pages 1158–1166, 2012.
[61] K. Shi and K. Ali. GetJar mobile application recommendations with very sparse datasets. In Proceedings of the 18th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, pages 204–212, 2012.
[62] P.-W. Soh, K.-H. Chen, J.-W. Huang, and H.-J. Chu. Spatial-temporal pattern analysis and prediction of air quality in taiwan. In proceedings of the 2017 International Conference on Ubi-Media Computing(UMedia), 2017.
[63] P. Sondhi, J.-M. Sun, H.-H. Tong, and C.-X. Zhai. SympGraph: A framework for mining clinical notes through symptom relation graphs. In Proceedings of the 18th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, pages 1167–1175, 2012.
[64] R.-B.-V. Subramanyam, A.-S. Rao, R. Karnati, S.-R. Suvvari, and D.-V.-N. Somayajulu. Mining closed sequential patterns in progressive databases. Journal of Information
[65] A. P. Tai, L. J. Mickley, and D. J. Jacob. Correlations between fine particulate matter (pm 2.5) and meteorological variables in the united states: Implications for the sensitivity of pm 2.5 to climate change. Atmospheric Environment, 44(32):3976–3984, 2010.
[66] J. Tang, S. Wu, and J.-M. Sun. Confluence: Conformity influence in large social networks. In Proceedings of the 19th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, pages 347–355, 2013.
[67] L.-A. Tang, X. Yu, Q.-Q. Gu, J.-W. Han, A. Leung, and T. L. Porta. Mining lines in the sand: On trajectory discovery from untrustworthy data in cyber-physical system. In Proceedings of the 19th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, pages 410–418, 2013.
[68] TWEPA. Air quality index historical data, http://taqm.epa.gov.tw/taqm/tw/YearlyDataDownload.aspx.
[69] TWEPA. PM2.5 index, http://taqm.epa.gov.tw/taqm/en/fpmi.htm, http://www.tnepb.gov.tw/AIR PM25.htm.
[70] X.-K. Wang and W.-Z. Lu. Seasonal variation of air pollution index: Hong kong case study. Chemosphere, 63(8):1261–1272, 2006.
[71] L.-Y. Wei, Y. Zheng, and W.-C. Peng. Constructing popular routes from uncertain trajectories. In Proceedings of the 18th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, pages 195–203, 2012.
[72] J. Yuan, Y. Zheng, and X. Xie. Discovering regions of different functions in a city using human mobility and pois. In Proceedings of the 18th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, pages 186–194, 2012.
[73] L. Yuan, Y.-L.Wang, P.-M. Thompson, V.-A. Narayan, and J.-P. Ye. Multi-source learning for joint analysis of incomplete multi-modality neuroimaging data. In Proceedings of the 18th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, pages 1149–1157, 2012.
[74] Y. Zheng, F. Liu, and H.-P. Hsieh. U-air: When urban air quality inference meets big data. In Proceedings of the 19th ACM SIGKDD international conference on Knowledge discovery and data mining, pages 1436–1444. ACM, 2013.
[75] Y. Zheng, X. Yi, M. Li, R. Li, Z. Shan, E. Chang, and T. Li. Forecasting fine-grained air quality based on big data. In Proceedings of the 21th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, KDD ’15, pages 2267–2276, New York, NY, USA, 2015. ACM.
[76] Y. Zheng, X. Yi, M. Li, R. Li, Z. Shan, E. Chang, and T. Li. Forecasting Fine-Grained Air Quality Based on Big Data Dataset, 2015. [Online; accessed 22-August-2016].