【摘要】：轉(zhuǎn)增和送股是近年來(lái)我國(guó)證券市場(chǎng)上市公司常用的股票紅利分配方式,然而對(duì)于轉(zhuǎn)增和送股對(duì)上市公司帶來(lái)的股價(jià)收益率影響卻是多個(gè)方面的。本文選取了2012年6月至2012年12月發(fā)生轉(zhuǎn)送股分紅的上市公司共計(jì)496家,通過(guò)數(shù)據(jù)挖掘的方法分別分析了各種因素對(duì)于轉(zhuǎn)送股分紅公告日后短期、中期和長(zhǎng)期收益率的影響,并運(yùn)用證券市場(chǎng)上的客觀規(guī)律來(lái)解釋數(shù)據(jù)挖掘方法得到的模型結(jié)果。 本文通過(guò)logistic回歸、分類樹和神經(jīng)網(wǎng)絡(luò)方法分別對(duì)三個(gè)時(shí)間跨度的收益率類別進(jìn)行模型擬合,結(jié)合實(shí)際對(duì)得到的模型參數(shù)進(jìn)行解釋,并對(duì)同一時(shí)間跨度的收益率類別所對(duì)應(yīng)的不同模型結(jié)果進(jìn)行比較和評(píng)價(jià),從而制定出較為合理的投資策略。通過(guò)上述工作,本文得出的結(jié)論分為方法和模型結(jié)果兩個(gè)部分。 在方法上,本文通過(guò)將數(shù)據(jù)挖掘得到的模型結(jié)果與現(xiàn)實(shí)證券市場(chǎng)上的客觀規(guī)律相對(duì)照,認(rèn)為兩者在相當(dāng)程度上存在對(duì)應(yīng)關(guān)系,即通過(guò)數(shù)據(jù)挖掘方法得到的各種因素對(duì)于轉(zhuǎn)送股公告日后收益率的影響關(guān)系,大多與現(xiàn)實(shí)市場(chǎng)上的客觀規(guī)律相符,因此本文認(rèn)為通過(guò)數(shù)據(jù)挖掘的方法研究證券市場(chǎng)中的一些現(xiàn)象以獲取有價(jià)值的信息并將其運(yùn)用于投資決策的做法是可行并且有意義的。 在模型結(jié)果方面,本文得到的關(guān)于轉(zhuǎn)送分紅公告后收益率變化結(jié)論如下： 1.分紅公告中存在送股的股票,絕大部分能夠在短期、中期和長(zhǎng)期都獲取正的相對(duì)收益率。 2.半年報(bào)收益率與收益率有正相關(guān)性。 3.市場(chǎng)對(duì)于高轉(zhuǎn)增股票所持的態(tài)度較為保守。 4.在公布轉(zhuǎn)增和送股信息的公司中,市場(chǎng)對(duì)于其市盈率的規(guī)模大小有一定的要求,市盈率在特定范圍內(nèi)的股票更有可能獲取正的收益。 與其他研究轉(zhuǎn)送股的文獻(xiàn)相比,本文的特點(diǎn)在于引進(jìn)了數(shù)據(jù)挖掘的方法,并將模型結(jié)果與實(shí)際規(guī)律相結(jié)合,制定出較為合理的投資策略和建議。因而,本文的研究工作不僅是數(shù)據(jù)挖掘方法在研究證券市場(chǎng)的應(yīng)用方面的一次探索,而且為我國(guó)證券市場(chǎng)的投資操作提供了一些有價(jià)值的參考。 然而,本文的研究工作仍然有相當(dāng)大改進(jìn)和提升空間,首先添加更多的輸入變量可以提升數(shù)據(jù)挖掘模型的預(yù)測(cè)性能；其次,本文的研究結(jié)果并沒(méi)有能夠揭示內(nèi)幕操作對(duì)于收益率的影響作用；最后,研究影響內(nèi)幕操作對(duì)于收益率的即時(shí)沖擊的因素仍有很多的工作要做。
[Abstract]:In recent years, transfer and share delivery are commonly used stock dividend distribution methods for listed companies in China's securities market. However, there are many aspects of the impact of stock price return on listed companies. In this paper, a total of 496 listed companies with dividends from June 2012 to December 2012 are selected, and the effects of various factors on the short-term, medium-and long-term returns after the announcement are analyzed by means of data mining. The objective laws in the securities market are used to explain the model results obtained by the data mining method. In this paper, logistic regression, classification tree and neural network methods are used to fit the return categories with three time span, and the model parameters are explained in combination with the actual situation. The results of different models corresponding to the return categories with the same time span are compared and evaluated, and a more reasonable investment strategy is worked out. Through the above work, the conclusions of this paper are divided into two parts: method and model results. In terms of method, this paper compares the results of the model obtained by data mining with the objective laws in the real securities market, and holds that there is a corresponding relationship between the two to a certain extent. That is, the influence of various factors obtained by data mining method on the future rate of return of the transfer stock announcement is mostly consistent with the objective law in the real market. Therefore, this paper holds that it is feasible and meaningful to study some phenomena in the securities market by data mining in order to obtain valuable information and apply it to investment decision-making. In terms of the results of the model, the results of this paper are as follows: 1. There are stocks in the dividend announcement, most of which can obtain a positive relative rate of return in the short, medium and long term. two銆，

本文編號(hào)：2491039