臺灣博碩士論文加值系統

對於全球發展中城市來說，提升運輸系統至關重要，為此，我們需要一個動態監控系統來監控通勤流量。使用這樣的方法有助於讓我們了解存在模式，開發移動性預測方法，並通過位置建議減少交通擁堵。此研究是基於現有的方法分析存在模式和從使用者的手機數據中提取他們的位置資訊。他們提出了一種新穎的方法可以根據個人的時空格局，準確的由通聯紀錄推斷通勤的起點和終點。而此研究提升了文獻中普遍採用的經驗分配法則的準確率。首先，此研究著重於使用我們提出的方法，Normalized Hourly Presence (NHP)，來了解使用者位置的存在模式。再來，我們使用FastICA演算法，一種更有效率的獨立成分分析方法來提取共同的存在模式
值得注意的是我們原本是使用主成分分析作為我們的分析方法，但效果並不理想，而這也是我們選擇使用主成分分析的原因，也因此得到較佳的結果。
由於我們由於我們使用的是大規模的資料集，因此時間複雜度可能會是一個問題，為了解決這個問題，我們使用了Improved-FastICA演算法。這個演算法提升了目標函數的收斂速度，使我們可以大幅的降低程式的執行時間，相較於原本使用的FastICA。最後，我們使用了非監督式的K-means演算法來提取出來自不同族群旅客的相似度為了標記各個群，我們分析並解釋了我們所得到的結果。
我們假設凌晨12:00到早上7:00和晚上7:00到半夜11:59為回家時間而早上8:00到晚上6:00為工作時間。因此，圖中顯示在回家時間有眾多人潮而在工作時間較無人潮聚集之處為家族群反之，則為工作族群。

In fast developing cities around the world, improving the transit systems is crucial, and for that we need a dynamic monitoring system to monitor the commuting flows. Using data from such system will allow the understanding of presence patterns, developing mobility prediction methods and also reducing traffic congestions by location recommendations.
This work is based on the profiling presence patterns and segmenting user locations from cell phone data, where the proposed novel method accurately infer the point of origin and destinations of commuting flows based on individual’s spatial-temporal patterns which is inferred from Call Detail Records (CDR). And this work improves the accuracies upon the heuristic assignment rules popularly adopted in the literature.
First, this work focuses on understand presence patterns at user locations, it uses the proposed metric called Normalized Hourly Presence (NHP).
Second, extract common presence patterns with the Independent Component Analysis (ICA), specially, with the FastICA algorithm. As we are working with large-scale dataset, the time complexity can be an issue, to face this problem, we used the Improved-FastICA algorithm for running time reduction. The algorithm improves the convergence speed of the objective function. we were able to reduce significantly the running time with the Improved-FastICA compared to the FastICA.
Finally, we use the K-means algorithm which is an unsupervised learning technique, to extract the similarities across locations for different groups of travelers.
To label the clusters, we analyze and interpret the graphs obtained. We assume that the time intervals [00:00am to 7:00am] and [7:00pm to 11:59pm] are home times, and [8:00am to 6:00pm] are work times. So, graphs that are showing high presence during home times and low presence during work times are considered to be home clusters, and those showing high presence during work times and low presence during home times are interpreted as work clusters.

摘要 ................................................................................................................................... I
ABSTRACT.......................................................................................................................... II Acknowledgments ............................................................................................................IV List of Tables ....................................................................................................................VI List of Figures ...................................................................................................................VI
1.
2. 3.
Introduction ........................................................................................................... - 1 -
1.1 Background ...................................................................................................................... - 3 - 1.2 Motivations...................................................................................................................... - 4 - Related work ......................................................................................................... - 5 - Methodology ......................................................................................................... - 6 - 3.1 Data description............................................................................................................... - 6 - 3.2 Conceptual framework..................................................................................................... - 7 - 3.3 Normalized hourly presence (NHP)................................................................................... - 8 - 3.4 Data Processing................................................................................................................ - 9 -
3.5 Dimensionality Reduction .............................................................................................. - 12 -
3.5.1 Principal Component Analysis ....................................................................................................... - 13 - 3.5.2 Independent Component Analysis ................................................................................................ - 14 -
3.6 Clustering....................................................................................................................... - 26 -
3.7 Summary........................................................................................................................ - 27 -
3.8 Running Environment..................................................................................................... - 28 -
3.8.1 Submitting Spark Applications....................................................................................................... - 30 -
Syntax of spark-submit: ....................................................................................................... - 30 -
Results analysis and comparisons......................................................................... - 33 -
4.1 PCA result vs FastICA result ............................................................................................ - 33 - 4.2 FastICA vs Improved FastICA .......................................................................................... - 35 -
4.2.1 FastICA and Improved-FastICA Time Comparison ......................................................................... - 36 -
4
5
References................................................................................................................... - 41 -

[1] JamesonLToole, SerdarColak, FahadAlhasoun, AlexandreEvsukoff, andMartaC Gonzalez. 2014. The path most travelled: mining road usage patterns from massive call data. arXiv preprint arXiv:1403.0636 (2014).
[2] Francesco Calabrese, Laura Ferrari, and Vincent D. Blondel. Urban sensing using mobile phone network data: A survey of research. ACM Comput. Surv., 47(2):25:1– 25:20, November 2014.
[3] Yu Zheng, Licia Capra, Ouri Wolfson, and Hai Yang. Urban computing: concepts, methodologies, and applications. ACM Transactions on Intelligent Systems and Technology (TIST), 5(3):38, 2014.
[4] Mi Diao, Yi Zhu, Joseph Ferreira, and Carlo Ratti. 2015. Inferring individual daily activities from mobile phone traces: A Boston example. Environment and Planning B: Planning and Design (2015). https://doi.org/10.1177/0265813515600896 arXiv:http://epb.sagepub.com/content/early/2015/09/15/0265813515600896.full.pd f+html
[5] YanLeng, AlejandroNoriega, Alex’Sandy’Pentland, IraWinder, NinaLutz, and Luis Alonso. 2016. Analysis of Tourism Dynamics and Special Events through Mobile Phone Metadata. arXiv preprint arXiv:1610.08342 (2016).
[6] IvaBojic, EmanueleMassaro, AlexanderBelyi, StanislavSobolevsky, andCarlo Ratti. 2015. Choosing the right home location definition method for the given dataset. CoRR abs/1510.03715 (2015). http://arxiv.org/abs/1510.03715
[7] Lauren Alexander, Shan Jiang, Mikel Murga, and Marta C Gonzá lez. 2015. Origin– destination trips by purpose and time of day inferred from mobile phone data. Transportation Research Part C: Emerging Technologies (2015).
[8] KevinSKung, KaelGreco, StanislavSobolevsky, andCarloRatti.2014.Exploring universal patterns in human home-work commuting from mobile phone data. (2014).
[9] Francesco Calabrese, Laura Ferrari, and Vincent D. Blondel. 2014. Urban Sensing Using Mobile Phone Network Data: A Survey of Research. ACM Comput. Surv. 47, 2, Article 25 (Nov. 2014), 20 pages. https://doi.org/10.1145/2655691
[10] Shan Jiang, Joseph Ferreira Jr, and Marta C Gonzá lez. [n. d.]. Activity-Based Human Mobility Patterns Inferred from Mobile Phone Data: A Case Study of Singapore.
[11] ChristianMSchneider, VitalyBelik, ThomasCouronné , ZbigniewSmoreda,and Marta C González. 2013. Unravelling daily human mobility motifs. Journal of The Royal Society Interface 10, 84 (2013), 20130246.
[12] Marta C Gonzalez, Cesar A Hidalgo, and Albert-Laszlo Barabasi. 2008. Under- standing individual human mobility patterns. nature 453, 7196 (2008), 779.
- 41 -
[13] Xin Lu, Erik Wetter, Nita Bharti, Andrew J Tatem, and Linus Bengtsson. 2013. Approaching the limit of predictability in human mobility. Scientific reports 3 (2013), srep02923.
[14] Yu Zheng, Licia Capra, Ouri Wolfson, and Hai Yang. Urban computing: concepts, methodologies, and applications. ACM Transactions on Intelligent Systems and Technology (TIST), 5(3):38, 2014.
[15] Michael Batty, Kay W Axhausen, Fosca Giannotti, Alexei Pozdnoukhov, Armando Bazzani, Monica Wachowicz, Georgios Ouzounis, and Yuval Portugali. Smart cities of the future. The European Physical Journal Special Topics, 214(1):481–518, 2012.
[16] Rein Ahas, Siiri Silm, Olle Järv, Erki Saluveer, and Margus Tiru. Using mobile positioning data to model locations meaningful to users of mobile phones. Journal of Urban Technology, 17(1):3–27, 2010.
[17] Francesco Calabrese, Massimo Colonna, Piero Lovisolo, Dario Parata, and Carlo Ratti. Real-time urban monitoring using cell phones: A case study in rome. Intelligent Transportation Systems, IEEE Transactions on, 12(1):141–151, 2011.
[18] Santi Phithakkitnukoon, Teerayut Horanont, Giusy Di Lorenzo, Ryosuke Shibasaki, and Carlo Ratti. Activity-aware map: Identifying human daily activity pattern using mobile phone data. In Human Behavior Understanding, pages 14–25. Springer, 2010.
[19] Shan Jiang, Gaston A Fiore, Yingxiang Yang, Joseph Ferreira Jr, Emilio Frazzoli, and Marta C González. A review of urban computing for mobile phone traces: current methods, challenges and opportunities. In Proceedings of the 2nd ACM SIGKDD International Workshop on Urban Computing, page 2. ACM, 2013.
[20] Kevin S Kung, Kael Greco, Stanislav Sobolevsky, and Carlo Ratti. Exploring universal patterns in human home-work commuting from mobile phone data. 2014.
[21] Lishan Sun, Liya Yao, Shuwei Wang, Jing Qiao, and Jian Rong. Properties analysis on travel intensity of land use patterns. Mathematical Problems in Engineering, 2014, 2014.
[22] Lauren Alexander, Shan Jiang, Mikel Murga, and Marta C González. Origin– destination trips by purpose and time of day inferred from mobile phone data. Transportation Research Part C: Emerging Technologies, 2015.
[23] Vincent D Blondel, Adeline Decuyper, and Gautier Krings. A survey of results on mobile phone datasets analysis. EPJ Data Science, 4(1):1–55, 2015.
[24] T.-W. Lee Independent component analysis
Independent Component Analysis, Springer (1998), pp. 27-66
[25] A. Hyvärinen, J. Karhunen, E. Oja
Independent Component Analysis, Vol. 46, John Wiley & Sons (2004)
- 42 -

[26] Xu Huang. An Improved FastICA Algorithm for Blind Signal Separation and its Application
[27] Yan Leng and Haris Koutsopoulos and Jinhua Zhao. 2018. Profiling presence patterns and segmenting user locations from cell phone data.cs.SI. arXiv.1805.12208