The Study on Forecasting Market Share of Scooter’s Brands Based on Deep Learning and Machine Learning Methods

• Main Authors: Shih-Chia Yeh, 葉時佳
• Other Authors: Ting Lie
• Format: Others
• Language: zh-TW
• Published: 2018
• Online Access: http://ndltd.ncl.edu.tw/handle/c6n6x3

碩士 === 元智大學 === 資訊管理學系 === 106 === Nowadays, the market is ever-changing, and companies need to understand their own positioning before they can seize any opportunity. Understanding the difference among competitors through market share is regarded as an important basis for corporate decision makers. Online forum media has tapped out the consumer's Internet word-of-mouth information to understand consumer's thoughts on the differen brands. This study uses web crawler to automatically collect three Taiwan scooter’s brand’s (KYMCO, SYM, YAMAHA) online word-of-mouth on, their three common engine displacements (100cc, 125cc, 150cc) models from the online forums. The word-of-mouth information is converted into monthly emotional indicators. Using machine learning and deep learning methods the data are the used to estimate market share and to predict sales volume The weights of the best-performing training modules are adopted to understand the factors affecting the market share of each brand.