考慮住宅周邊環境之房價預測分析模型及服務系統 ─ 以台北市與新北市為例

House Price Prediction Model with Consideration of Neighborhood Features – A Case Study of Taiwan

謝子宸(Zih-Chen Hsieh)

指導教授 ： 袁賢銘

國立交通大學/資訊學院/網路工程研究所/碩士(2018年)

摘要

現有台灣房價研究及預測服務多半以房屋特性做為參數，對於房屋週遭的因素研究較少。然而，影響房價因素不僅包含房子本身的特性，房價也會受到房屋週遭環境以及時間的影響。
本論文以台北市以及新北市的公開房價資料做為研究對象，藉由分析房屋環境因素(neighborhood attributes)對房價之影響，定義出有效的環境參數，最後提出一個結合房屋本身、環境因素、以及時間趨勢性的房價預測模型。由實驗的結果可以得知，考慮到環境因素的房價預測模型可以有較好的預測效果，此外，與靜態預測模型相比，時間序列模型具有較佳的預測能力。
　　最後，基於提出的模型，我們也提出的房價預測服務，讓使用者能夠利用此服務得到一個有效的預測結果作為買房的參考。實驗的結果也顯示出服務的預測能力以及可靠度與預測模型一致。

關鍵字

房價預測 ； 房價分析 ； 時間序列模型 ； 房屋周遭參數 ； 環境參數

並列摘要

Nowadays, most research and services of house price prediction in Taiwan focus on house characteristics and seldom take neighborhood/environmental features into consideration. However, attributes that affect house price not only contain house attributes, but also include neighborhood attributes and temporal trend of house price.
This thesis used open data of real estate transaction data in Taipei City and New Taipei City, analyzed the effect of neighborhood attributes on house price to define a set of effective neighborhood attributes, and finally proposed a model that combine house attributes, neighborhood attributes, and temporal trend of house price to make house price prediction. In experiments, results indicated that neighborhood-considering prediction model has a better performance in prediction loss. And, in comparison to static model, the time-series model has a better prediction capability.
Finally, based on the proposed model, this thesis also proposed a house price prediction web system that for help users to estimate budget when buying houses. Also, the experiment showed that service has same prediction performance and reliability as well as model.

並列關鍵字

house price prediction ； house price analysis ； time-series model ； neighborhood attributes ； environmental attributes

參考文獻

[1] P.-Y. Lai, "Applying the Artificial Neural Network in Computer-assisted Mass Appraisal," JOURNAL OF HOUSING STUDIES, vol. 16, pp. 43-65, 2007.

[2] Y. Feng and K. Jones, "Comparing Multilevel Modelling and Artificial Neural Networks in House Price Prediction," presented at the Spatial Data Mining and Geographical Knowledge Services (ICSDM)

2015 2nd IEEE International Conference, 8-10 July 2015,

[3] V. Limsombunchai, "House price prediction: Hedonic price model vs.artificial neural network," presented at the NZARES Conference, 2004.

[4] F.-Y. Chen and I.-J. Young, "The Accuracy of Valuation: A Comparison of Appraisers and Mass Appraisal," JOURNAL OF HOUSING STUDIES, vol. 16, pp. 67-84, 2007.