以亂數基礎分類法與螞蟻聚類演算法探討電子零組件及電腦週邊在填權息之實證研究

長期投資者對股票是否能填權息一直有深度的疑慮，哪些是股票財務屬性影響填權息的重要因子? 股票的財務結構、配息的比例和股價常常是影響股票填權息的重要因子，本研究以電子零組件及電腦週邊類股為研究對象，以亂數基礎分類法進行屬性刪減及螞蟻聚類聚類演算法分析來了解影響股票填權息之影響關係。研究期間以2014年6月至2015年6月雅虎股市電子零組件及電腦週邊類股共計65筆，以其十一項財務屬性進行資料庫統計，並利用亂數基礎分類法與螞蟻聚類聚類演算法來分析股類進而篩選決定投資標的。本研究之目的在探討股市中那些屬性因子影響填權息，以半年內填權息區分(a)使用全部屬性(b)亂數基礎分類法篩選重要屬性進行模型建立，另以一年內填權息區分(c)使用全部屬性(d)亂數基礎分類法篩選重要屬性進行模型建立，四種不同的案例分析研判最具影響填權息因素。本研究預期能推估影響股票填權率相關屬性，從中決定選擇股票準確率。

關鍵字

亂數基礎分類法；螞蟻聚類演算法；填權及填息

並列摘要

The long-term investor has a basic question arise in his mind: which kind of financial attributes influence the return of ex-dividend effectiveness? The financial structure, ex-dividend rate, major ratio of inventory and stock price are often the important factors affecting stock ex-dividend. In this study, an electronic components are used for the analysis to understand the return of ex-dividend. There are 65 datasets with 11 financial and stock-price related attributes employed in this study. The study period is June 2014 to June 2015. Then, the entropy-based classification will used as an attribute extraction tools to cut and select the important influenced attributes. Then, the particle swarm optimization is used as a classifier to discern the stock data which can be successfully return of ex-dividend. There are four different cases are designed as: (a)the three months period with original data on stock ex-dividend (b) in half a year period with extraction data on stock ex-dividend (c) the a whole year period with original data on stock ex-dividend(d) the a whole year period with extraction data on stock ex-dividend. The outcomes are made for understanding the relations on financial attribute and return of ex-dividend.