臺灣博碩士論文加值系統

近年來，空氣污染日益嚴重。 人們非常關注PM2.5對人體的影響。 PM2.5對人體有很多影響，包括心血管和呼吸系統，甚至肺癌。預測空氣品質尤為重要。本文在大數據平台下開發了三種機器學習方法，決策樹回歸（DTR），梯度增強樹回歸（GBTR）和支持向量回歸（SVR），建立了空氣品質預測模型。通過訓練多維數據，在台灣的67個不同地點建立了空氣品質預測模型。 通過結合SVR的ARIMA補值方法，我們建立了一個8小時的空氣品質預測模型。 通過ETL架構獲取實時數據，我們可以即時預測PM2.5的數值。

In recent year, air pollution is getting worse. People pay lots of attention to the impact of PM2.5 on the human body. PM2.5 has many affects to the human body, including cardiovascular and respiratory systems, and even lung cancer. The forecasting of air quality is particularly important. In this paper, three machine learning methods, decision tree regression(DTR),gradient boosted tree regression(GBTR) and support vector regression(SVR), are developed under the big data platform to establish an air quality forecasting model. By training multi-dimensional data, air quality forecasting model are built for 67 different locations in Taiwan. Through the ARIMA complement method combined with SVR, we built an 8-hour air quality forecasting model. Obtaining real-time data through the ETL architecture, and we can instantly forecast the value of PM2.5.

ABSTRACT III
目錄 IV
圖目錄 VI
表目錄 VIII
第一章 緒論 1
1.1現況與問題 1
1.2目標與貢獻 4
1.3論文架構 5
第二章 背景與相關研究 6
2.1大數據平台 6
2.1.1 Apache Spark 6
2.1.2 MongoDB 8
2.2演算法介紹 9
2.3相關研究 13
第三章 機器學習演算法 15
3.1決策樹回歸 15
3.2梯度增強樹回歸 19
3.3支援向量回歸 23
3.4整合移動平均自回歸模型 26
第四章 系統架構與流程 30
4.1即時資料處理 30
4.1.1 ETL架構 30
4.2模型訓練 31
4.3空氣品質預測 34
4.3.1一小時空氣品質預測 34
4.3.2八小時空氣品質預測 37
第五章 實作與分析 40
5.1實驗環境 40
5.2系統實作 40
5.3精準度評估 52
5.4結果分析及比較 54
5.5 模型訓練時間比較 64
5.6 討論 65
第六章 結論與未來目標 67
參考文獻 68

[1] Qinghua Sun, MD, PhD; Aixia Wang, BS; Ximei Jin, BS “Long-term air pollution exposure and acceleration of atherosclerosis and vascular inflammation in an animal model”. JAMA. 2005;294(23):3003-3010
[2] Michelle L. Block and Lilian Calderón-Garcidueñas “ Air pollution: mechanisms of neuroinflammation and CNS disease”. Volume 32, Issue 9, September 2009, Pages 506-516
[3] Sermin Genc, Zeynep Zadeoglulari, Stefan H. Fuss, and Kursad Genc “The Adverse Effects of Air Pollution on the Nervous System”. Journal of Toxicology, Volume 2012, Article ID 782462, 23 pages
[4] https://taqm.epa.gov.tw/taqm/tw/default.aspx
[5]http://opendata.cwb.gov.tw/index;jsessionid=416B2ABE26E7F596D55EAF0D75CDE0B6
[6] Yong Liu a, Huaicheng Guo, Guozhu Mao, Pingjian Yang “A Bayesian hierarchical model for urban air quality prediction under uncertainty”. Atmospheric Environment 42 (2008) 8464–8469
[7] Qingping Zhou, Haiyan Jiang , Jianzhou Wang, Jianling Zhou “A hybrid model for PM2.5 forecasting based on ensemble empirical mode decomposition and a general regression neural network”. Science of the Total Environment 496 (2014) 264–274
[8] Ana Russo, Frank Raischel, Pedro G. Lind “Air quality prediction using optimal neural networks with stochastic variables”. Atmospheric Environment 79 (2013) 822 - 830
[9] M.A. Elangasinghe, N. Singhal, K.N. Dirks, J.A. Salmond, S. Samarasinghe, “Complex time series analysis of PM10 and PM2.5 for a coastal site using artificial neural network modelling and k-means clustering”. Atmospheric Environment 94 (2014) 106 - 116
[10] Sankar Ganesh S, Sri Harsha Modali, Soumith Reddy Palreddy, Dr. Arulmozhivarman P, “Forecasting air quality index using regression models: A case study on Delhi and Houston”. 2017 International Conference on Trends in Electronics and Informatics (ICEI), 248 – 254
[11] Ke Hu, Vijay Sivaraman, Hari Bhrugubanda, Shiying Kang, Ashfaqur Rahman, “SVR based dense air pollution estimation model using static and wireless sensor network”. 2016 IEEE SENSORS, Pages: 1 - 3
[12] Aorong Luo, Xiaoli Li, Yang Li, Jiangeng Li, “Application of accurate online support vector regression in atmospheric SO2 concentration prediction”. 2018 Chinese Control And Decision Conference (CCDC), Pages: 6274 - 6279
[13] Fangchen Su, Yinliang Xu, Xiaoying Tang, “Short-and mid-term load forecasting using machine learning models”. 2017 China International Electrical and Energy Conference (CIEEC), Pages: 406 - 411
[14] Mihaela Oprea, Marian Popescu, Sanda Florentina Mihalache, “A neural network based model for PM2.5 air pollutant forecasting”. 2016 20th International Conference on System Theory, Control and Computing (ICSTCC), Pages: 776 - 781
[15] K. Siwek, S. Osowski, M. Sowinski, “Neural predictor ensemble for accurate forecasting of PM10 pollution”. The 2010 International Joint Conference on Neural Networks (IJCNN), Pages: 1 – 7
[16] S. H. Yu, Y. S. Koo, E. Y. Ha, H. Y. Kwon, “PM-10 Forecasting Using Neural Networks Model’’. 2008 International Conference on Computational Intelligence for Modelling Control & Automation, Pages: 426 – 429
[17] Giovanni Raimondo, Alfonso Montuori, Walter Moniaci, Eros Pasero, Esben Almkvist, “Data-driven models to forecast PM10 concentration’’. 2007 International Joint Conference on Neural Networks, Pages: 190 - 194
[18] Fani A. Tzima, Kostas D. Karatzas, Pericles A. Mitkas, Stavros Karathanasis, “Using data-mining techniques for PM10 forecasting in the metropolitan area of Thessaloniki, Greece’’. 2007 International Joint Conference on Neural Networks, Pages: 2752 - 2757