以Char演算法協同特徵化公司財務表現; Characterizing Cooperate Financial Performance with Char Algorithm

NCUIR > School of Management at National Central University > Graduate Institute of Industrial Management > Electronic Thesis & Dissertation > Item 987654321/11151

Please use this identifier to cite or link to this item: http://ir.lib.ncu.edu.tw/handle/987654321/11151

Title:	以Char演算法協同特徵化公司財務表現;Characterizing Cooperate Financial Performance with Char Algorithm
Authors:	劉蕙瑜;Hwei-Yu Liu
Contributors:	工業管理研究所
Keywords:	財務報表;小波;資料採礦;Wavelet;Financial Statement;Data Mining
Date:	2005-05-27
Issue Date:	2009-09-22 14:15:03 (UTC+8)
Publisher:	國立中央大學圖書館
Abstract:	現代投資人遇到的問題並非缺少可分析的資料，而是面對這麼多在網路上唾手可得的財務資料，根本不知從合分析起。因此本篇論文的研究動機即是想幫投資人從表現良好的上市電子公司之財務報表中，歸納出既有的特色，找出特徵化會計科目或是財務比率的組合,進而建議投資人，在閱讀財務報表時，可以專注於哪些特定少數的特徵化屬性，即可以使用較少的力氣，得到較多的資訊內涵。此篇論文，在前置過程中,依序使用滑動視窗，處裡上市公司資料長短不依的問題；接著，應用小波技術解決時間序列的問題；之後使用階層分層技術，來離散化資料，使之資料型態可以適用於Char演算法。最後由資產報酬率觀點，我們選擇了107家表現良好的台灣上市電子公司進行實驗，以Char演算法找出財務報表中的特徵規則，得到結果發現,在總共88的會計科目和財務比率中，成長率這個群組的屬性,公司良好表現有很大的關聯性。因此建議投資人在分析財務報表時，可以多注意這些特徵規則。 In the perspective of investments, they desire to use the less financial variables to anticipate the most performance information. Every human being is increasingly faced with unmanageable amounts of financial data; hence, data mining or knowledge discovery apparently affects all of us. In this study of mining performance of company, we attempt to summarize the stronger characteristic rules of fundamental analysis using 81 financial statement variables. To address this problem, we proposed an effective method, a Char Algorithm, to automatically produce characteristic rules to describe the major characteristics of data in a table is proposed. To fit the data type of Char Algorithm, we proceed many steps to preprocess source data of financial statement from 2001-2003. In the first step, data compression, we adapt wavelet methods to preprocess time series data of several attributes from financial statement from 2001-2003. After data to be compressed by wavelet technique, the second step, sliding window, processes in order to increase the amount of virtual data. Thirdly, we use cluster method to do data discretization process categorizing data to fit the discrete data type. It is a difficult task to construct a concept tree to describe the financial statement. In contrast to traditional Attribute Oriented Induction methods, the algorithm, named as Char Algorithm, does not need a concept tree and only requires setting a desired coverage threshold to generate a minimal set of characteristic rules to describe the given dataset. We develop a formal framework for financial data to adapt Char Algorithm and afford advisements to investors to extract characteristic rules, rapidly. It is also our observation that the dimension of growth rate is significant in circumstance of generalizing good performance companies.
Appears in Collections:	[Graduate Institute of Industrial Management] Electronic Thesis & Dissertation

Files in This Item:

File	Size	Format

社群 sharing

Loading...