整合語意分析與平行決策超平面技術於營運表現之預測

CCUR > College of International Studies and Foreign Languages > English Program of Global Business > Thesis > Item 987654321/48497

Please use this identifier to cite or link to this item: https://irlib.pccu.edu.tw/handle/987654321/48497

Title:	整合語意分析與平行決策超平面技術於營運表現之預測 Integration of Linguistic Cues and Parallel Decision Surface for Performance Forecasting
Authors:	周家秀
Contributors:	全球商務碩士學位學程碩士班
Keywords:	Operating performance forecasting
Date:	2020
Issue Date:	2020-08-26 14:05:53 (UTC+8)
Abstract:	The financial crisis that escalated over a decade ago had severe devastating effects that lasted for years. It left behind in its trail both economic and social side effects far from being favorable. While the phenomenon inclined many researchers to devise prediction models which act as security for anticipating future financial turmoil, there has been little attention brought upon corporate operating performance, an important business factor that has been widely deemed to be one of the main causes of financial crisis. Considering the research gap, this study introduces a synthesized structure wherein qualitative and quantitative approaches are fused together to build a hybrid operating performance forecasting model. It is generally considered that warning signs for financial distress are recognizable, provided that, investors are equipped with a proficient and reliable forecasting system. Through the proposed model, managers and investors alike will be able to anticipate the problem and make changes where necessary. The focal point of the quantitative information is on publicly listed companies in the semiconductor industry, using financial ratios as predictors of operating performance. However, since merely utilizing financial ratios is distinguished as insufficient and results in an unconvincing prediction model, this study supplements numerical analysis with textual analysis. Quantitative data is derived from computational linguistics in the form of annual report readability. The research covers the years spanning 2016 – 2018 and focuses on the semiconductor industry as it comprises some of Taiwan’s largest companies and is considered to be a major contributor to the economy. Results are then logged into support vector machines (SVMs), a type of AI-based technique, grounded on the statistical learning theory and structural risk minimization (SRM) principle. The SVMs method has demonstrated its superior generalization ability in several forecasting tasks in the past few years. The presented model supplements numerical study with textual analysis in order to generate a more robust and reliable performance forecasting method that serves as a better gauge for future financial crises.
Appears in Collections:	[English Program of Global Business] Thesis

Files in This Item:

File	Description	Size	Format
index.html		0Kb	HTML	166	View/Open

View Licence

Loading...