基于模糊支持向量机的上市公司财务困境预测

支持向量机(SVM)已经成功地应用于财务困境预测问题的研究,且已证明优于多元线性判别分析(MDA)、逻辑回归(Logistic regression)和神经网络(NN)等方法.然而,传统SVM使用结构风险最小化的原则,这样可能导致错误分类的经验风险升高,特别是当样本点与最优超平面十分接近的时候,这种误分类的经验风险显著升高.另外,传统SVM还存在过拟合问题,所以对数据集中的外点或噪声十分敏感.因此,采用模糊支持向量机(FSVM)算法来改进上述不足.首先,建立一个适当的成员模型用于对整个数据集的模糊处理;然后通过外点侦察方法(ODM)来发现外点,其中ODM集成了模糊C-均值算法(Fuzzy C-mean algorithm)和无监督神经网络中的自组织映射(SOM).最后,为主体集和外点集中的样本点分配不同的权值.还将FSVM应用于上市公司财务困境预测的实证研究,实证结果表明FSVM与传统SVM相比,FSVM 能较好的解决经验风险升高和过度拟合问题,确实降低了外点的影响并提高了分类器的分类准确率.     

I-FIFOWA算子及其在群决策中的应用

对模糊数直觉模糊信息的集结算子进行了进一步研究,引入了模糊数直觉模糊的一些运算法则、模糊数直觉模糊的得分函数,并基于这些运算法则,提出了一些新算子:模糊数直觉模糊加权平均(FIFWA)算子、模糊数直觉模糊有序加权平均(FIFOWA)算子和诱导的模糊数直觉模糊有序加权平均(I-FIFOWA)算子,然后分析了它们的性质.针对属性权重和专家权重为实数,属性取值为模糊数直觉模糊数的多属性群决策问题,给出了一种基于FIFWA算子和I-FIFOWA算子的模糊数直觉模糊多属性群决策方法.最后,进行了实例分析,说明了该方法的实用性和有效性.     

一种基于信息熵与K均值迭代模型的模糊聚类算法

本文提出了基于信息熵和K均值算法混合迭代模糊聚类的细分模型,解决了模糊聚类的原型初始化参数问题。将信息熵和K均值算法引入模糊聚类中进行分析,并结合测试样本数据进行实际分析,与传统方法相比,取得了较好的效果。     

模糊库存模型中模糊数置信区间的研究

本文对模糊库存模型中的模糊数的置信区间进行研究,利用历史数据估计出库存变量的概率分布,在此基础上通过置信度和置信区间估计模糊数。最后利用该方法估计最佳缺陷率模糊数,并将该方法用于模糊库存模型,合理估计最佳订货量。     

半模糊超球支持向量机多类分类方法研究

针对支持向量机在大类别模式分类中存在的问题,提出了一种基于半模糊核聚类的超球支持向量机分类方法.该方法首先利用半模糊核聚类方法对样本进行预处理,完成边缘样本的选取,进而以所选样本为训练样本进行超球支持向量机训练,从而有效提高分类器的性能.实验表明,该方法比标准支持向量机多类分类方法具有更高的速度和精度.     

信息不完全确定的区域产品模糊区间聚类方法

区域产品分类与选择是区域经济发展中最重要和最基础的工作.在产品选择与分类时,需要确定指标权系数和分类阈值等参数,这在实际应用中是比较困难的.针对这种情况,提出了一种信息不完全确定的区域产品模糊区间聚类方法.该方法构建了指标权系数信息不完全确定的最优模糊区间聚类模型,利用遗传算法和改进的FCM算法联合求解所得优化模型,得到指标权系数、最优聚类中心和最优划分,进而确定各产品所属类别.最后将该方法应用于某区域的产品分类和主导产品的确定中,实例计算说明该方法的可行性和有效性.     

运用数据挖掘技术提高网络银行客户量

通过结合运用模糊均值聚类和感知机进行网络银行目标营销,首先利用模糊c均值方法(FCM)找出频繁使用网络银行服务的客户,再进一步用这些客户样本对一个多层感知机(MLP)进行训练,最后利用训练好的感知机对银行的非网络银行注册客户进行分类,找出潜在的网上银行服务频繁使用者,银行即可对这些客户进行重点推广网络银行服务。     

模糊C聚类综合评价法在企业财务信用评级中的应用

本文针对当前企业财务信用问题,提出了基于模糊C均值聚类和综合评价相结合的财务等级分类方法。探讨了聚类分析方法在财务信用分类中的应用,包括模糊聚类方法和根据聚类中心进行综合评价研究两个方面。采用实际的数据进行了实证分析,表明该方法可根据实际需要获取很好的评价分类效果,而且根据聚类中心进行评价,能更清楚地反映信用等级。     

基于SODM的多分类器融合及其在客户分类中的应用

目前大多数客户分类研究常采用单一的分类模型.引入多分类器融合方法,并将其与自组织数据挖掘理论(SODM)相结合,提出了基于SODM的选择性融合算法(SBSF).SBSF从全部基分类器的分类结果组成的初始模型集出发,由上一层模型两两组合产生新的待选模型,用最小二乘法来估计融合权重,而用外准则来评价和筛选中间候选模型,直到根据终止法则找到最优复杂度的融合模型.在15个UCI数据集上的实验结果显示,与单一的分类模型以及常用的多数投票法、贝叶斯方法、遗传算法等5种融合方法相比,SBSF具有更高的分类精度.进一步地,在信用卡数据集"german"上的客户分类实验表明,SBSF能自适应地从基分类器池中选择一个适当的基分类器子集进行融合,从而提高客户分类的精度.     

基于广义模糊数的软件成本加权CBR估算研究

在软件项目开发过程中,准确估算出软件成本在提高软件质量和保障软件成功开发方面起到重要支撑作用。针对软件项目历史数据库中部分属性在项目开发初期难以给予精确数值(仅仅能给出模糊数),而已有软件成本估算模型不能很好地处理模糊信息的问题,本文在基于案例推理模型(CBR)基础上集成广义模糊数,提出了基于广义模糊数的CBR模型。使用基于广义模糊数的相似度度量方法代替传统CBR模型中采用的欧式距离等相似度度量方法,采用模糊C均值聚类(FCM)方法将已有软件项目历史数据库中的精确数值进行模糊化处理,以匹配新项目中的模糊数。进一步采用粒子群算法(PSO)来优化属性的权重,构建基于广义模糊数的加权CBR模型。最终在实验中采用Desharnais数据来检验构建模型的有效性。实证结果表明,在与常用的欧式距离CBR模型相比,构建的基于广义模糊数的加权CBR模型能有效提高估算精度,采用PSO优化属性权重能提高模型的估算精度。     

Class‐Based Storage with a Finite Number of Items: Using More Classes is not Always Better

Class‐based storage is widely studied in the literature and applied in practice. It divides all stored items into a number of classes according to their turnover. A class of items with higher turnover is allocated to a region closer to the warehouse depot. In the literature, it has been shown that the use of more storage classes leads to a shorter travel time for storing and retrieving items. A basic assumption in this literature is that the required storage space for all items equals their average inventory level, which is valid only if an infinite number of items can be stored in each storage region. This study revisits class‐based storage by considering each storage space to contain only a finite number of items. We develop a travel time model and an algorithm that can be used for determining the optimal number and boundaries of storage classes in warehouses. Different from the conventional research, our findings illustrate that commonly a small number of classes is optimal. In addition, we find the travel time is fairly insensitive to the number of storage classes in a wide range around the optimum. This suggests that a manager can select a near‐optimal number of storage classes in an easy way and need not be worried about the impact of storage‐class reconfigurations. We validate our findings for various cases, including different ABC‐demand curves, space‐sharing factors, number of items, storage rack shapes, discrete storage locations, and stochastic item demand.     

大规模灾害中基于聚类的医疗物资联合运送优化

应急响应中,往往出现救援物资供应节点与需求节点距离太远、关键道路损毁导致难以及时通过车辆运送物资到灾区等情景,此时直升飞机逐渐被用来运送关键应急资源（如医疗物资及医护人员）。然而,大规模灾害中难以使用直升飞机运送医疗物资到每个医疗救助点,通常考虑灾民的聚集性选择一定数量的应急中转点,以接收直升飞机运送的医疗物资,之后采用车辆运送物资到其覆盖的医疗救助点。针对该问题,提出一种基于聚类的两阶段医疗物资联合运输方法：第一阶段根据医疗救助点分布,采用模糊C-均值算法（FCM）进行应急中转点选择和医疗救助点划分,并针对FCM划分中存在的剩余容量不均衡问题,考虑容量约束提出一种改进划分方法（FCMwCC）,构建“直升飞机-车辆”医疗物资联合运输网络结构;第二阶段建立基于聚类的运送路线优化模型,确定从应急中转点到医疗救助点的具体运送路线。数值实验验证了提出方法和算法的有效性。     

Inventory Rationing for Multiple Class Demand under Continuous Review

We consider how a firm should ration inventory to multiple classes in a stochastic demand environment with partial, class‐dependent backlogging where the firm incurs a fixed setup cost when ordering from its supplier. We present an infinite‐horizon, average cost criterion Markov decision problem formulation for the case with zero lead times. We provide an algorithm that determines the optimal rationing policy, and show how to find the optimal base‐stock reorder policy. Numerical studies indicate that the optimal policy is similar to that given by the equivalent deterministic problem and relies on tracking both the current inventory and the rate that backorder costs are accumulating. Our study of the case of non‐zero lead time shows that a heuristic combining the optimal, zero lead time policy with an allocation policy based on a single‐period profit management problem is effective.     

一种基于PIM与核方法的模糊聚类新算法

本文针对模糊C均值聚类在大数据量时收敛较慢以及不能对多种数据结构有效聚类的缺点,结合PIM算法与核方法提出了一种新的高效聚类算法———KPIM算法,并从理论上证明了该算法的收敛性.最后利用标准实验数据IRIS数据集测试,结果表明KPIM算法在保证收敛速度的同时,聚类效果更有效.     

RATIONALIZATION OF THE SUPPLIER BASE IN THE PRESENCE OF YIELD UNCERTAINTY

An important aspect of supply chain management is the optimal configuration of the supplier base. We develop a model to determine optimal lot sizes and the optimal number of suppliers when the yield of the product delivered from each supplier is random. While small orders from a large number of suppliers can reduce yield uncertainty, fixed costs associated with each supplier provide a penalty for having too many suppliers. This is the key tradeoff addressed by our model. We look at the cases when the suppliers are identical as well as nonidentical.     

OPTIMAL BUFFER ALLOCATION IN ASYNCHRONOUS CYCLIC MIXED-MODEL ASSEMBLY LINES

This paper considers a problem of optimal buffer allocation in cyclic asynchronous mixedmodel assembly lines with deterministic processing times. An analytical model is used to provide new insights into properties of optimal buffer allocation, that is, a buffer configuration that guarantees the highest possible throughput rate on the assembly line with a minimum number of buffers. Optimal buffer configuration is characterized, and an efficient algorithm to find such a configuration is developed. The approach proposed in this paper also provides insights on how to allocate a given number of buffers to workstations on the assembly line to maximize the throughput rate.     

在线拍卖中的“托”出价研究

本文分析了独立的私人估价环境下最优保留价格与竞买人数之间的关系,纠正了最优保留价格与竞买人数无关的传统观点.在此基础上,讨论了卖者可能递交"托"出价的条件以及拍卖中只剩下一个竞买人时卖者的最优"托"出价.     

缺陷汽车召回所需安排的最优维修店数目

讨论了缺陷汽车产品制造商在汽车召回事件中如何安排最优维修店数目。假设汽车制造商在召回期限内维修几乎全部召回汽车,车主逗留时间和维修时间都服从指数分布,并且假设维修人员会根据库存水平(车位)调整维修速度。选取制造商安排组织的维修店数目为决策变量,利用有限源排队理论建立了成本控制模型。搜索算法确定了最优的维修店数目,使得期望总成本最小。     

Reliability analysis of man–machine systems using fuzzy cognitive mapping with genetic tuning

This article offers a method for analyzing the reliability of a man–machine system (MMS) and ranking of influencing factors based on a fuzzy cognitive map (FCM). The ranking of influencing factors is analogous to the ranking of system elements the probabilistic theory of reliability. To approximate the dependence of “influencing factors—reliability,” the relationship of variable increments is used, which ensures the sensitivity of the reliability level to variations in the levels of influencing factors. The novelty of the method lies in the fact that the expert values of the weights of the FCM graph edges (arcs) are adjusted based on the results of observations using a genetic algorithm. The algorithm's chromosomes are generated from the intervals of acceptable values of edge weights, and the selection criterion is the sum of squares of deviations of the reliability simulation results from observations. The method is illustrated by the example of a multifactor analysis of the reliability of the “driver–car–road” system. It is shown that the FCM adjustment reduces the discrepancy between the reliability forecast and observations almost in half. Possible applications of the method can be complex systems with vaguely defined structures whose reliability depends very much on interrelated factors measured expertly.     

Near-optimal control policy for loss networks

In this paper, the phenomenon of the optimal management of requests of service in general networks is formulated as a control problem for a finite number of multiserver loss queues with Markovian routing. This type of problem may arise in a wide range of fields, e.g., manufacturing industries, storage facilities, computer networks, and communication systems. Using inductive approach of dynamic programming, the optimal admission control can be induced to be the functions of the number of requested service in progress. However, for large-scale network, the computational burden to find optimal control policy may be infeasible due to its involvement of the states for all stations in the networks. Hence, the idea of bottleneck modeling is borrowed to compute the near-optimal admission control policy. We reduced the scale of loss network and decreased the difference between the original and reduced models by making compensation for system parameters. A novel method is proposed in this paper to compute the compensation. Numerical results show that the near-optimal control policy demonstrates close performance to the optimal policy.