臺灣博碩士論文加值系統

如今，wifi的普及度越來越高，其在各物聯網絡之間扮演著重要的腳色，為了滿足使用者的需求，必須提高wifi的效率。 此研究在國立虎尾科技大學裡進行，蒐集學校設置在校園各處的wifi，利用三種生存分析方法預測未來wifi使用情況的有效性，並透過各處wifi蒐集到使用者使用wifi的歷史數據，其中以使用者使用wifi的時間、日期和使用者名稱等資料來做研究分析。
生存分析是對於感興趣事件相關的數據分析，本論文實施的三種生存分析方法為Kaplan-Meier估計、Cox比例風險模型和隨機生存林，所提出的方法實現了大約30％的錯誤率，這表明生存分析方法獲得了滿意的預測結果，通過理解協方差的連結並完成有效的決策，該方法可以使用於改善任何組織中的WiFi網路組織。

Nowadays WiFi plays a significant role help access the Internet. In order to reach the high expectation of wireless users, the performance of the wireless network should be improved. This work presents the effectiveness of adopting three survival analysis approaches to predict the WiFi usage in the future and the understanding of covariance affect WiFi usage such as date, time, and user, by introducing dataset of WiFi usage historical. The study took place in National Formosa University in Taiwan.
Survival analysis is the analysis of data associated times to event of interest. The three survival analysis methods implemented in this paper are Kaplan-Meier estimator, Cox Proportional Hazards Model, and Random Survival Forest. The proposed approaches achieve an error rate of approximately 11% which shows that survival analysis approach gains a satisfy prediction result. This approach can be adapted for improving WiFi network organization in the future by understanding the connection of covariance and accomplishing an effective decision.

Chinese Abstract ......................... i
English Abstract .........................ii
Acknowledgement .........................iii
Table of Contents.........................iv
List of Tables............................ v
List of Figures...........................vi
List of Abbreviations....................vii
Chapter 1 Introduction.....................1
1.1 Research Motivation....................1
1.2 Thesis Objectives......................2
1.3 Thesis Outline.........................3
Chapter 2 Background and Related Works.....4
2.1 Survival analysis......................4
2.2 Censored data..........................5
2.3 Survival and Hazard function...........6
2.4 Survival Analysis Study................8
2.5 Related Survival Approaches............9
Chapter 3 Study of Methodology............12
3.1 Kaplan-Meier (KM).....................12
3.1.1 Kaplan-Meier in RStudio.............13
3.2 Cox Proportional Hazards (Cox PH).....13
3.2.1 Cox Proportional Hazards in RStudio.14
3.3 Random Survival Forest (RSF)..........15
3.3.1 Random Survival Forest in RStudio...16
Chapter 4 Results and Discussion..........17
4.1 The datasets..........................17
4.2 Kaplan Meier Survival.................18
4.3 Cox Proportional Hazards Model........21
4.4 Random Survival Forest................23
4.5 Comparison and predictive performance.25
Chapter 5 Conclusion and Future work......27
Reference.................................19
Extended Abstract

[1] Dan Pan, “Analysis of Wi-Fi performance data for a Wi-Fi throughput prediction approach”, KTH, 2017.
[2] iPass Corporate, “iPass Mobile Professional Report 2016”, iPass company, 2016.
[3] Christiana Kartsonaki, “Survival analysis”, University of Oxford, Oxford, UK, 2016. [4] Melinda Mills, “Introducing Survival and Event History Analysis”, SAGE Publications, 2011
[5] Kleinbaum D.G., “Survival Analysis, a self learning text”, Springer-Verlag, 1996. [6] Jutarat Kositnitikul, Ji-Han Jiang, “Survival Model for WiFi Usage Forecasting in National Formosa University ”, National Formosa University, Taiwan , 2018.
[7] Eman Alhasawi, “Survival Analysis Approaches for Prostate Cancer”, Laurentian University, Sudbury, Ontario, Canada, 2015.
[8] Ajay Byanjankar, “Predict Credit Risk in Peer-to-Peer Lending with Survival Analysis”, Facultly of Social Sciences, Business and Economicsm Abo Akademi University, Turku, Finland, 2017.
[9] Summer M. Husband, Jason Roberts, Randstad Sourceright Houston, TX , 2017.
[10] A Rizkiana, H. Sari, P. Hardjomijojo, B. Prihartono, T. Yudhistira, “Analyzing the Impact of Investor Sentiment in Sociial Media to Stock Return:Survival Analysis Approach”,Bandung Institute of Technology, Bandung, Indonesia, 2017
[11] Hosmer D.W., Lemeshow S., and May S., Applied Survival Analysis: Regression Modeling of Timeto- Event Data, Wiley, 2008.
[12] T. Smith and B. Smith, “Kaplan Meier And Cox Proportional Hazards Modeling: Hands On Survival Analysis”, SAS® Users Group International Proc. Seattle, Washington, 2003. [13] Cox, D. R. “Regression Models and Life-Tables.” Journal of the Royal Statistical Society. Series B (Methodological), vol. 34, no. 2, 1972, pp. 187–220. JSTOR, JSTOR, www.jstor.org/stable/2985181.
[14] F. Louzada, V.G. Cancho, M.R. Oliveira, and B. Yiqi, “Modeling Time to Default on a Personal Loan Portfolio In Presence of Disproportionate Hazard Rates”, J Stat App Pro., vol. 3(3), pp.295-305, 2014.
[15] Hemant Ishwaran, Udaya B. Kogalur, Eugene H. Blackstone and Michael S. Lauer, “Random Survival forests”, Cleveland Clinic, Columbia University, 2018.
[16] Ulla B. Mogensen, Hemant Ishwaran , A. Gerds , “Evaluating Random Forests for Survival Analysis Using Prediction Error Curves”, Department of Biostatistics, University of Copenhagen, 2012.
[17] R Core Team, “R: A Language and Environment for Statistical Computing”, R Foundation for Statistical Computing, Vienna, Austria, 2014. [18] Terry M Therneau and Thomas Lumley. http://CRAN.R-project.org/package=survival (core), 2009.
[19] Ishwaran and Kogalur. http://CRAN.R-project.org/package=randomForestSRC, 2014.
[20] John Ehrlinger, “ggRandomForests: Exploring Random Forest Survival”, Microsoft, 2016.