多获胜节点SOM及其在股票分析中的应用

为了增强自组织映射（self-organizing map, SOM）网络的动态竞争和聚类能力,提高解的精度,在无监督的SOM神经网络的基础上,通过拓广获胜节点的数量,改进网络中的邻域函数和连接权函数等方法,提出具有多获胜节点的SOM模型.为了避免多个输入样本映射到同一个输出节点,还提出了禁忌映射的方法.为了验证所提出的方法的有效性,以股票的聚类分析为实例,对该方法进行了检验.通过对每股收益、每股净资产、净资产收益率、每股经营性现金流量及净利润等5项反映上市公司综合盈利能力的财务指标进行了模拟实验,所得的数值结果表明,在标准SOM及所提出的几种多获胜节点SOM网络模型中,具有双获胜节点（SOM with 2 winners, SOM2W）的网络模型获得了最好的聚类效果.结合实验结果对网络模型的进一步分析也表明,SOM2W的聚类能力优于标准SOM及其他网络模型.该模型为股票的分析和选择提供了一种可行的途径,在金融领域具有潜在的应用价值.     

基于GEP和CPN网络的文本聚类算法

针对传统的方法很难做到根据输入向量的实际分布来设置Kohonen层各神经元对应的权向量的状况,因其会影响文本的聚类质量,所以利用人工神经网络和基因表达式编程(GEP)的互补优势,通过利用GEP在组合优化的方法进行对CPN网络中Kohonen层的联接权向量的优化,提出了一种基于GEP和CPN网络的文本聚类算法(GCTCA)。通过实验结果表明了该算法在文本聚类上的有效性与优越性。     

基于模糊kohonen聚类网络的改进算法

模糊kohonen聚类神经网络将模糊隶属度概念应用于一般Kohonen聚类网络的学习和更新策略中,改善了Kohonen聚类网络的性能,是一种快速有效的聚类网络.但在死神经元的处理和收敛速度上还有改进的空间.为了能使网络更好应用于海量数据的聚类问题,对模糊Kohonen聚类网络算法在输出神经元的模糊偏置度、侧抑制模糊隶属度和加权系数提出了三方面改进.同时,对改进的模糊Kohonen聚类网络的有效性进行实例仿真,仿真结果体现了改进算法能有效避免死神经元的出现和提高了网络的聚类速度.     

IWO-Kohonen聚类算法在IDS中的应用

针对Kohonen神经网络模型网络入侵聚类正确率较低的问题,将入侵杂草优化(IWO)算法与Kohonen神经网络相结合,提出IWO-Kohonen聚类算法。利用IWO算法优化Kohonen神经网络的初始权值,训练Kohonen神经网络模型得到最优值。使用IWO算法增强算法的搜索能力,提高聚类正确率,并加快算法的收敛速度。实验结果表明,该算法与模糊聚类算法和广义神经网络聚类算法相比,分类正确率较高;与蚂蚁聚类算法和模糊C均值聚类算法相比,网络入侵检测率较高,误报率较低。     

一种统计降维和Kohonen网络相结合的文本聚类方法

提出了一种基于词条互信息（WMI）值的统计降维和Kohonen网络（SOFM网）相结合的文本聚类方法,WMI值的方法侧重考虑文本特征项之间的互信息进行降维,可提高特征选择的效率,并使其更趋实用化。采用Kohonen网络进行文本聚类,其学习率函数是随时间单调下降的退火函数,实验结果表明了这种结合方法较一般的降维方法得到的聚类结果具有较高的聚类精度。     

基于Eviews的上市公司股票价格的影响因素分析

在金融市场中,股票由于其收益性、流动性等特点吸引了众多投资者的目光,但需要注意的是影响股价的因素有很多,尤其值得关注的是企业本身的财务状况.为了更好地分析上市公司的财务状况与股价之间的关系,从某证券交易所的上市公司中随机抽取200家(不包含有ST公司),取2011-2020年间每个公司的经营现金流、每股收益、每股净资产这3项指标以及公司的股价,利用Eviews进行回归分析,并给出结论.     

Kohonen SOFM神经网络及其演化研究

Kohonen SOFM神经网络广泛地应用于模式聚类、模式识别、拓扑不变性映射等方面。从KohonenSOFM神经网络结构和聚类算法入手，对其演化网络进行了比较分析，并从聚类算法性能的角度给予了综述。最后针对网络结构和算法的不足，指出了需进一步研究的方向。     

基于模式聚类和遗传算法的文本特征提取方法

采用模式聚类和遗传算法进行文本特征提取,并用Kohonen网络进行分类。模式聚类可以有效降低文本特征的维数,使得特征从几千维降为几百维。但几百维的维数对Kohonen网络来说仍然太高,因此采用遗传算法在此基础上继续降维。实验结果表明,这两种方法结合可以极大地降低文本的维数,并能提高分类准确率。     

基于Kohonen神经网络聚类方法在遥感分类中的比较

设计完成和比较了基于Kohonen自组织网络的Kohonen聚类网络（Kohonen Clustering Network, KCN）、模糊Kohonen聚类网络（Fluzzy KCN, FKCN）和基于进化规划的Kohonen聚类网络（Evalutionary Programming based KCN, EPKCN）三种聚类算法在遥感土地利用/覆盖分类中的应用。结果表明三种非监督学习方法在进行遥感土地利用/覆盖分类过程中，在分类性能上有显著差异。EPKCN分类目视效果最好，单次迭代的速度最快；FKCN总的收敛速度最快；而按遥感土地利用/覆盖分类要求而言，EPKCN方法在三种分类方法中效果最好，因此可采用该算法进行遥感土地利用/覆盖的非参数分类。     

基于一维Kohonen网络的遥感影像分类方法研究

分析了Kohonen网络的训练模式和聚类特性,选用规模相对较小的一维Kohonen网络,并调整网络输出层的规模和邻域形状,优化网络结构;同时根据多光谱遥感影像中地物波谱曲线特征,通过不同波段组合、波段权重系数调整等方法对输入数据进行预处理,使该方法更适用于多光谱遥感影像分类和专题提取.本文以浙江省绍兴地区多光谱遥感影像分类为例,研究结果表明使用改进后的分类方法可以有效提高分类精度.     

财务困境预测模型与仿真研究

公司财务困境受到决策者、市场、经济和政策多种因素影响,针对传统预测方法预测精度低缺陷,提出了一种贝叶斯判别分析的财务困境预测方法。首先选用了反映公司财务状况的24个指标作判别因子,建立了公司财务困境的贝叶斯判别分析模型,然后采用85个上市公司的实际数据作为学习样本建立贝叶斯判别函数,以交差确认估计法对判别准则进行评价,以验证模型的有效性,最后利用判别函数对5个待评价公司进行预测,得到判别函数值,进行仿真。结果表明,采用贝叶斯判别分析模型提高了公司财务困境的预测精度,是一种有效的财务困境预测方法。     

Network mining: Applications to business data

This research addresses the problem of analyzing the temporal dynamics of business organizations. In particular, we concentrate on inferring the related businesses, i.e., are there groups of companies that are highly correlated through some measurement (metric)? We argue that business relationships derived from general literature (i.e., newspaper articles, news items etc.) may help us create a network of related companies (business networks). On the other hand, relative movement of stock prices can give us an indication of related companies (asset graphs). We also expect to see some relationships between these two kinds of networks. We adapt the asset graph construction approach from the literature for our asset graph implementations, and then, define our methodology for business network construction. Finally, an introduction to the exploration of some relationships between the asset graphs and business networks is presented.     

AdaBoost ensemble for financial distress prediction: An empirical comparison with data from Chinese listed companies

Due to the important role of financial distress prediction (FDP) for enterprises, it is crucial to improve the accuracy of FDP model. In recent years, classifier ensemble has shown promising advantage over single classifier, but the study on classifier ensemble methods for FDP is still not comprehensive enough and leaves to be further explored. This paper constructs AdaBoost ensemble respectively with single attribute test (SAT) and decision tree (DT) for FDP, and empirically compares them with single DT and support vector machine (SVM). After designing the framework of AdaBoost ensemble method for FDP, the article describes AdaBoost algorithm as well as SAT and DT algorithm in detail, which is followed by the combination mechanism of multiple classifiers. On the initial sample of 692 Chinese listed companies and 41 financial ratios, 30 times of holdout experiments are carried out for FDP respectively one year, two years, and three years in advance. In terms of experimental results, AdaBoost ensemble with SAT outperforms AdaBoost ensemble with DT, single DT classifier and single SVM classifier. As a conclusion, the choice of weak learner is crucial to the performance of AdaBoost ensemble, and AdaBoost ensemble with SAT is more suitable for FDP of Chinese listed companies.     

Does stock repurchase declaration affect stock price? Differences between the electrics industry and other industries

This study is to discuss the impact of stock repurchase declaration and purpose of repurchase on the stock price in the backdrop of listed companies on Taiwan’s stock market. Event Study Method is employed to discuss stock price fluctuations while GARCH (Generalized Autoregressive Conditional Heteroscedasticity) is applied to estimate the Market Model regressive coefficients. The samples consisted of companies declaring first stock repurchase are selected from August 9, 2000 to December 31, 2005 with a precondition that all the companies shall be listed ones 150 days prior to declaration. The study results reveal that companies from other industries have considerably bigger average CAR than companies of the electrics industry before and after the declaration of stock repurchase. Companies with application purpose of “maintain stockholders’ equities and corporate credit” have considerably bigger average CAR than companies with application purpose of “transferring stocks to employees”. In industries other than electrics, companies with application purpose of “maintain stockholders’ equities and corporate credit” have bigger accumulated abnormal return response than companies with application purpose of “transferring stocks to employees”. In case of “maintain stockholders’ equities and corporate credit” as the application purpose of stock repurchase, companies from industries other than electrics have relatively higher average CAR response. The empirical study results can serve as a reference for the listed company management and to related academic studies.     

基于集成学习的上市公司高送转预测实证研究

我国证券市场中高送转题材股备受中小投资者的追捧,但市场中也存在着借高送转概念炒作的乱象,如何利用上市公司的财务数据挖掘真正有潜力的股票无疑具有重要意义。采用2?158家制造业上市公司7年的财务指标作为研究数据,利用采样、特征选择以及集成学习算法构建上市公司高送转预测模型并进行实证研究。结果显示：采样和特征选择方法均能有效提高集成预测模型的性能;相较于数据集中的冗余信息,数据不平衡问题对模型预测准确率的影响更显著;ADASYN+mRMR+XGBoost组合模型取得了最好的预测结果,高送转样本的分类准确率达到84.96%,建议投资者优先选用该组合模型对上市公司的高送转情况进行预测。     

Wavelets Analysis on Structural Model for Default Prediction

In recent years, to improve predictive ability of corporate defaults has become an important problem. In this paper, regarding on characteristics of listed companies, we sampled 100 companies according to industry types, constructed wavelet structural model, experimented with wavelet decomposition proceeds to get low frequency and high frequency sequence, built the prediction model for both sequences, and then using the prediction of future returns to reconstruct predictive returns, thus avoiding accumulated prediction process with earnings volatility of time series model, therefore enhanced the precision of default prediction. Finally we compared wavelet structural model with time series structural model based on the predictive default distance of China’s listed companies.     

基于交叉覆盖算法的股票预测

针对股票预测的特点,选择了对上市公司股票走势有重要影响的相关数据并对其进行测试,为了避免传统的神经网络分类方法(如BP算法)的学习过程收敛速度慢、网络性能差、可能存在局部极小值等弊端,文中使用可以避免这些弊端并且具有良好分类功能的交叉覆盖算法对上市公司股票走势进行预测,结果表明预测的精度明显高于采用BP算法等传统神经网络分类方法的测试结果,且速度快,结果的可解释性强,预测达到了让人满意的效果。     

基于支持向量机的股票预测

针对股票预测的特点,选择对上市公司股票走势有重要影响的相关数据进行测试。为了避免传统的预测算法(如BP算法)的一些弊端,使用可以避免这些弊端并且具有良好分类功能的支持向量机对该上市公司股票走势进行预测。测试表明预测的精度明显高于采用BP算法等传统神经网络分类方法的测试结果,预测达到了让人满意的效果。     

Predicting business failure using forward ranking-order case-based reasoning

With the rapid development of business computing for Chinese listed companies, it is focused on to use case-based reasoning (CBR) in business failure prediction (BFP). Ranking-order case-based reasoning (RCBR) uses ranking-order information among cases to calculate similarity in the framework of k-nearest neighbor. RCBR is sensitive to the choice of features, meaning that optimal features can help it produce better performance. In this research, we attempt to use wrapper approach to find the optimal feature subset for RCBR in BFP. Forward feature selection method and RCBR are combined to construct a new method, namely forward RCBR (FRCBR). The combination is implemented by combining forward feature selection with RCBR as a wrapper module. Hold out method is used to assessing the performance of the classifier. Empirical data were collected from Chinese listed companies in the Shenzhen Stock Exchange and Shanghai Stock Exchange. We employed the standalone RCBR, the classical CBR with Euclidean metric as its heart, the inductive CBR, the two statistical methods of logistic regression and multivariate discriminate analysis (MDA), and support vector machines to make comparisons. For comparative methods, stepwise MDA was employed to select optimal feature subset. Empirical results indicated that FRCBR can produce dominating performance in short-term BFP of Chinese listed companies.