Use Normalized Google Distance Formula to Predict Public Opinion Intention and the Result of Poll Election

碩士 === 輔仁大學 === 統計資訊學系應用統計碩士在職專班 === 104 === This research exploited the characteristics of the Internet to retrieve the public opinions by NGD formula, and provided more time-sensitive and lower cost than the traditional method. This research used two different electoral systems, such as regional l...

Full description

Bibliographic Details
Main Authors: KAO,JUI-HUI, 高瑞徽
Other Authors: TU,YI-NING
Format: Others
Language:zh-TW
Published: 2016
Online Access:http://ndltd.ncl.edu.tw/handle/78456943220001603330
Description
Summary:碩士 === 輔仁大學 === 統計資訊學系應用統計碩士在職專班 === 104 === This research exploited the characteristics of the Internet to retrieve the public opinions by NGD formula, and provided more time-sensitive and lower cost than the traditional method. This research used two different electoral systems, such as regional legislative elections and municipal council elections respectively, and using the different time conditions to predict the election results. The experimental results showed that using NGD formula to predict the election results will reach the accuracy of 76.11% - 84.93% regardless of the type of electoral system. If take Taiwan as an example, the eastern region and outer islands in legislative elections can have 100% in accuracy. Therefore, the proposed method using NGD formula to predict polls and election results in the future. If there are time constraints or insufficient funds situation occurs, investigators can retrieve the Internet of public opinion polls first to get instant and relatively low cost information.