题名 |
應用CRISP-DM建立台灣區國產車銷售量預測模型 |
并列篇名 |
CRISP-DM to the Forecasting Model of Car Sales in Taiwan |
DOI |
10.6338/JDA.201304_8(2).0007 |
作者 |
梁耕華(Keng-Hua Liang);張芳仁(Fang-Jen Chang);邱俊誠(Chun-Cheng Chiu) |
关键词 |
汽車 ; 資料採礦 ; 類神經網路 ; 隨機森林 ; 時間序列 ; 廻歸分析 ; Cars ; Data Mining ; ANN ; Random Forests ; SVR ; Regression |
期刊名称 |
Journal of Data Analysis |
卷期/出版年月 |
8卷2期(2013 / 04 / 01) |
页次 |
129 - 155 |
内容语文 |
繁體中文 |
中文摘要 |
汽車產業一直是國家經濟發展的重要指標,隨著時代的進步,台灣地區的小客車數量已經由民國八十年的2,541,364輛成長至民國一百年的5,937,337輛,以台灣約兩千三百萬人口不到四個人就擁有一輛小客車,然而在民國九十八年台灣經濟成長率呈現負成長,在國產車的銷售數量也降為186,005輛,甚至不到尖峰時期民國94年447,558輛的二分之一,而如今景氣在慢慢回溫之下,本研究希望建立汽車銷售量預測模型以供相關產業或是政府單位以供決策之用。本研究以使用支援向量廻歸、類神經網路、時間序列、廻歸分析和隨機森林並以CRISP-DM建模流程建立台灣區國產車銷售量預測模型,分別對國產小客車銷售量、國產商用車銷售量、裕隆汽車、國瑞汽車和中華汽車做銷售量預測模型。本研究結果發現,不同的車型或者不同廠商在不同的月份對所適用的模型也有所不同。 |
英文摘要 |
The auto industry has been an important indicator of national economic development. As time goes by, there are 5,937,337 vehicles in Taiwan in 2011. But there are only 2,541,364 in 1991. There are about 2,300 million people in Taiwan, and it means less than four people have one car. However, the economic growth rate of Taiwan is the negative number and the number of car sales dropped to 186,005. Even not the half of the number 447,558 in 2005. But now, economic becomes better and better in Taiwan. The study hope to build a car sales forecasting model for the related industries or government agencies for making decisions.The study use SVR, neural network, Regression, Time Series and Random Forests, and use the CRISP-DM modeling process to forecast car sales with two kinds of cars and three kinds of auto manufacturers in Taiwan. The results of this study, different models or different vendors in different months, the applicable models are different. |
主题分类 |
基礎與應用科學 >
資訊科學 基礎與應用科學 > 統計 社會科學 > 管理學 |
参考文献 |
|