隨著桃園機場捷運的開通，桃園市與台北市的連結變得越來越緊密，甚至很有可能成為下一個台北或併入大台北都會區。桃園市的房價表現出前所未有的潛力，引起了廣泛的重視。因此本研究應用kNN模型去預測桃園地區不動產價格。首先將不動產總價劃分等級，來進行kNN的實際分類應用。 通過將不動產交易數據分別進行9:1與8:2的隨機切分，進行kNN模型的建立，並進行樣本內、外預測。實證分析發現，樣本內、外預測呈現高度一致性，以及高達0.76的預測精度和高達0.7的kappa值。另外我們還進行了時間上的穩健性測試，抽取每一年的樣本量，通過8:2的切分進行kNN模型的建立，並做In-sample和Out-sample的預測，實證結果肯定本模型的穩健性。

With the opening of Taoyuan Airport Express, the connection between Taoyuan City and Taipei City has become more and more close. It is possible that Taoyuan City will become the next Taipei or merge into the Greater Taipei Metropolitan Area. The house price of Taoyuan City has shown unprecedented potential and attracted extensive attention. Therefore, this study uses the kNN model to predict the real estate price in Taoyuan area. Firstly, the total real estate price is classified into different grades, and the actual classification and application of kNN are carried out. By randomly dividing the real estate transaction data into 9:1 and 8:2, the kNN model is established and the in-sample and out-sample predictions are made. Empirical analysis shows that the in-sample and out-of-sample predictions are highly consistent, and the prediction accuracy is as high as 0.76 and the kappa value is as high as 0.7. In addition, we also conducted a time robustness test, extracted the sample size of each year, established the kNN model through 8:2 segmentation, and made in-sample and Out-sample predictions. The empirical results confirm the robustness of the model.

第一章、緒論 ┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈ 1
第一節、研究背景┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈1
第二節、研究動機┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈3
第三節、研究架構┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈4
第二章 文獻回顧與相關理論┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈5
第一節、kNN應用於分類與台灣房價研究相關文獻回顧┈┈┈┈┈5
第二節、kNN原理說明┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈11
第三章 研究方法┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈ 13
第一節、資料來源┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈ 13
第二節、資料預處理┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈ 14
第三節、變數定義┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈ 15
第四節、kNN模型建立┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈20
第五節、k的取值┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈21
第六節、統計檢驗說明┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈ 22
第四章 實證結果與分析┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈ 23
第一節、交叉驗證結果┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈ 23
第二節、9:1與8:1隨機切分結果┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈26
第三節、按每年時間切分結果┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈ 29
第五章 結論與建議 ┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈35
參考文獻┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈36

沈育生，林秋瑾，2012，不同人工神經網路架構在不動產大量估價之應用與比較，台灣土地研究，第15卷，第1期，頁1-29.

林祖嘉，林素菁，1993，台灣地區環境品質與公共設施對房價與房租影響之分析，住宅學報，第1期，頁21-45.

蘇炳華，史秉璋，熊宗镥，1993，kappa檢驗的確切概率計算法，上海第二醫科大學學報，Vol.13，No.1，1993

鄧斌，邵培基，劉名武，夏國思，2010，基於模糊測度KNN的多維度數據分類算法，系統工程System s Engineering，Vol，28，No，3，Mar，2010

林素菁，2002，台灣地區特征性房價函數估計係數不一致性問題之探討，2002年中華民國住宅學會第二十一屆年會論文集，pp：268-277

張怡文，江穎慧，張金鶚，2009，分量迴歸在大量估價模型之應用－非典型住宅估價之改進，都市與計劃（民國九十八年），第三十六卷，第三期，pp281-304

連紹成(Shao-Cheng Lien)，2013，捷運規劃及動工時期對房價的影響 -以桃園地區透天住宅為例，國立中央大學產業經濟研究所碩士論文

武康平，胡諜，2011，房地產價格在宏觀經濟中的加速器作用研究，Chinese Journal of Management Science，Vol.19，No.1

洪得洋，林祖嘉，1999，台北市捷運系統與道路寬度對房屋價格影響之研究，住宅學報，第8期，pp.47-67。

M. Rahmatian，L. Cockerill，2004，Airport noise and residential housing valuation in southern
California: A hedonic pricing approach，International Journal of Environmental Science & Technology，1(1)，pp.17-25

Thanh N. Tran，RonWehrens，Lutgarde M.C. Buydens，2006，KNN-kernel density-based clustering for high-dimensional multivariate data，Computational Statistics & Data Analysis 51 (2006) 513 – 525

Andreas Mense，Konstantin A. Kholodilin，2014，Noise expextations and house prices: the reaction of property prices to an airport expansion，The Annals of Regional Science，(2014) 52:763–797

Austin Boyle，Charles Barrilleaux，Daniel Scheller，2014，Does Walkability Influence Housing Prices? SOCIAL SCIENCE QUARTERLY, Volume 95, Number 3, September 2014

Yu hong Yang，2007， Consistency of cross validation for comparing regression procedures，The Annals of Statistics，2007，Vol.35，No.6，2450-2473

Po-Ling Yu，M.D.，2007，Nuclear Factor-Kappa B Expression with Clinicopathological Factor in Colorectal Adenocarcinoma，Taipei City Med J 2007；4（3）：212-218

Donald E.Farrar，Robert R.Glauber，1967，Multicollinearity in Regression Analysis: The Problem Revisited，The Review of Economics and Statistics，Vol.49，No.1，pp92-107

David A.Stanley，Anthony Bonner，Zhaolei Zhang，Zineng Yuan，Renqiang Min，2009，A Deep Non-Linear Feature Mapping for Large-Margin kNN Classification，2009 Ninth IEEE International，Conference on Data Mining 2009

Francisco J.Pulgar，Franciso Charte，Antonio J.Rivera，Maria J.del jesus,AEkNN：An AutoEncoder kNN-based classifier with buit-in dimensionality reduction，nternational Journal of Computational Intelligence Systems,12（1），Feb 2018

Jiunn-Lin Wu，I-Jing Li，2010，A SOM-based dimensionality reduction method for KNN classifiers，International Conference on System Science & Engineering 19 August 2010

Shichao Zhang，Xuelong Li，Ming Zong，Xiaofeng Zhu，Debo Cheng，2017，Learning k for kNN Classification，ACM Transactions on Intelligent Systems and Technology，Vol. 8，No. 3，Article 43，Publication date:：January 2017.

Zhang S，Li X，ZongM，Zhu X，Wang R，2017，Efficient kNN Classification With Different Numbers of Nearest Neighbors，IEEE Trans Neural Netw Learn Syst.，2018 May，29(5):1774-1785. doi: 10.1109/TNNLS.2017.2673241. Epub 2017 Apr 12.

Cannisi，Dylan，Yuan，Bo，2016，Design Space Exploration for K-Nearest Neighbors Classification Using Stochastic Computing，2016 IEEE International Workshop on Signal Processing Systems (SiPS)

Wangmeng Zuo，David Zhang，Kuanquan Wang，2008，On kernel difference-weighted k-nearest neighbor classification，Pattern Analysis and Applications（2008）11:247-257 DOI 10.1007/s10044-007-0100-z

Vassilios Petridis，Vassilis G. Kaburlasos，2003，FINkNN: A Fuzzy Interval Number k-Nearest Neighbor Classifier for Prediction of Sugar Production from Populations of Samples，Journal of Machine Learning Research，4 (2003) 17-37

Patidar D，Shah B.C，Mishra M.R，2016，Performance analysis of K Nearest Neighbors image classifier with different wavelet features，International Conference on Green Computing Communication & Electrical Engineering，2014/03/01

Muhammad Waqar Aslam，Zhechen Zhu，Asoke K Nandi，2012，Automatic Modulation Classification Using Combination of Genetic Programming and KNN，IEEE Transactions on Wireless Communications 11(8):2742-2750 • August 2012

Muhammad Bilal，Huma Israr，Muhammad Shahid，Amin Khan，2016，Sentiment classification of Roman-Urdu opinions using Naïve Bayesian, Decision Tree and KNN classification techniques，Journal of King Saud University - Computer and Information Sciences，Volume 28, Issue 3, July 2016, Pages 330-344

Nenad Tomasev，Milos Radovanovic，Dunja Mladenic，Mirjana Ivanovic，2011，A probabilistic approach to nearest-neighbor classification: Naive hubness Bayesian kNN，Acm International Conference on Information & Knowledge Management 10/24/2011

Mahmoud Mejdoub， Chokri Ben Amar，2011，Classification improvement of local feature vectors over the KNN algorithm，Multimedia Tools and Applications，May 2013, Volume 64, Issue 1, pp 197–218

G. Donald Jud & Daniel T. Winkler，2006，The Announcement Effect of an Airport Expansion on House Prices，The Journal of Real Estate Finance and Economics，(2006) 33: 91–103，DOI 10.1007/s11146-006-8943-4

Claudio A. Agostini，Gaston Palmucci，2008，The Anticipated Capitalisation Effect of a New Metro Line on Housing Prices，Fiscal Studies 29(2):233-256 • June 2008

Meizhi Xiong，Xusong Jun，Ouyang Jun，2011，Spatio-temporal Impact Effects of Guangzhou Metro 3rd Line on Housing Prices，Scientia Geographica Sinica， 2011-07

Shengxiao Li，Luoye Chen，Pengjun Zhao，2017，The impact of metro services on housing prices: a case study from Beijing，Transportation（2017），pp 1–27

Guan, J.and A. Levitan，2008，An Adaptive Neuro-Fuzzy Inference System Based Approach to Real Estate Property Assessment，Journal of Real Estate Research，30(4 )，pp.395-421.

Chin-Oh Chang，Shu-Mei Chen，2018，Dilemma of Housing Demand in Taiwan，International Real Estate Review，2018 Vol.21 No.3: pp.397–418

Leo Breiman，1996，Bagging predictors，Machine Learning，24,pp.123-140（1996）

Sandrine, Dudoit，Jane，Fridlyand，2003，Bagging to improve the accuracy of a clustering procedure，Bioinformatics (Oxford, England)，2003，Vol.19 No.9 pp.100-119