基於實價登錄的房價模型研究

Abstract

近年來世界各國吹起開放政府資料的熱潮，我國內政部也將實價登錄制度推上檯面，以促進台灣房地產交易資訊的透明化。本篇主要工作在於建立包含實價登錄資料在內的多源的房地產交易資料收集系統，並利用資料視覺化技術，透過簡單易懂的圖表呈現，讓使用者能迅速了解複雜的房市交易資料並掌握房市走向。最後依據實價登錄資料建立房價預測模型估算房價。 我們收集了101年8月到102年9月的實價登錄資料以及有巢氏代售房屋資料。視覺化方面以架設網站的方式做資料的呈現，除了長條圖、折線圖，還用了堆疊圖、散布圖等多樣化圖表呈現台灣房市。房價模型建構則依據實價登錄系統中之交易資料為依據，利用SPSS統計分析軟體建立線性迴歸的房價模型以進行房價推算。我們以坪數、屋齡、格局、房屋種類等資訊為參數，透過半對數模型對台北市各區的房屋價格的估算，提供大眾買屋或賣屋的價格參考依據。

In recent years, open data is a trend in many country to access public-interesting data. In the meanwhile, the Ministry of the Interior, Taiwan has been promoting the Real Price Registration System to make the real estate transaction data public. In this work, we would like to build a system to consistently real estate price data from various sources for future research works, visualize the collected data for easy understanding, and develop skills to model the house price market. In this work, we built a database to collect house price data, including the real estate transaction data obtained from the Real Price Registration System (from August 2012 to September 2013), and the real estate pricing data periodically crawled from the U-Trust website. A web site was created to demonstrate the collected data by various visualization techniques, including bar charts, line charts, stack charts, and other diversified graphs to reveal the trends of the housing market in Taiwan. House price models were constructed by SPSS linear regression based on the transactions records from the Real Price Registration System. The parameters used in the models include the floor number, building age, pattern, house types, etc. We developed the price models for the districts of Taipei City.