基于多智能集成学习的中短期电煤价格预测

为扩大电力市场交易量与下调市场电价,需要提升电煤价格预测的可靠性与准确性.为此本文提出了多智能集成学习的中短期电煤价格预测方法.首先,阐述了Stacking集成学习的结构和原理;然后,介绍了数种智能电煤价格的预测模型,并通过算例证明了不同单智能模型对数据的感知能力存在差异性;进而,通过比较单智能模型预测结果的差异值均差,筛选出预测性能优异并且数据感知角度差异性明显的智能模型组.为了充分发挥个模型感知能力差异性的优势,利用Stacking融合各模型,得到一种适用于电煤价格滚动预测的集成模型.最后,通过滚动预测2019至2020年的电煤价格,对集成模型的有效性进行验证.     

基于集成模型的移动应用广告转化率预测

移动应用广告是互联网广告市场中一种主动的广告形式,它能够分析用户的兴趣爱好,并投其所好,精准投放广告,从而提高用户体验,为广告平台与广告主带来巨大的收益.因此,预测移动应用广告的转化率已成为一个非常重要的研究方向.本文以逻辑回归和两个梯度提升树模型为基础,使用堆叠和平均的集成思想,提出了两种集成模型--SXL和BLLX模型,解决了传统预测模型能力有限,无法精准预测转化率的问题.在腾讯2017社交广告比赛的数据集上的实验结果表明,SXL和BLLX两种模型能够有效地提高广告转化率的预测结果.     

产品差异度与双寡头厂商行为的实验分析

我们沿着Bertrand理论渊源构建模型,并以此为基础设计实验,在实验室环境中检验产品差异度与双寡头厂商的行为。论文首先分析模型的理论值,其次在实验市场中进行了设置检验。实验结果表明:(1)产品差异度越小,双寡头厂商的博弈平均价格就越低,该价格在重复博弈纳什均衡区间的中间价格附近波动;(2)博弈期数越多,博弈双方的冲突反而越大,这与俗定理预测相悖;(3)对背叛行为的惩罚具有普遍性,惩罚策略主要是针锋相对,惩罚强度随产品差异度减小而增强;(4)产品差异度对共谋的稳定性影响不显著,这与理论预测相悖;(5)折现与平均价格及共谋的关系模糊。     

基于模仿创造的网络流行语传播模型及仿真研究

随着互联网和智能移动终端的发展,研究网络流行语的传播过程和发展趋势对于网络营销和广告文案的创作具有重要的意义。基于SIR传染病模型,综合考虑网民对网络流行语模仿再创造的行为特点,构建新型的网络流行语传播模型,利用神经网络技术结合流行语时序数据对模型进行参数反演,并分别以流行语“佛系”和“确认过眼神”为例进行验证。结果表明,用户的模仿再创造行为是网络流行语传播中后期的主要驱动力;相较SIR模型,本模型着重考虑了网民对流行语的创新行为并运用参数反演方法,其预测准确度更高,模型拟合值与真实数据相比误差更小。进而可以为营销和广告创意人员提供有益的借鉴,并通过预测其发展趋势可对舆论进行及时分析和引导。     

基于CMMI的软件过程性能模型

在高成熟度组织的CMMI(能力成熟度模型集成)实施过程中,正确的建立并使用过程性能模型对有效的策划、预测、监督和控制组织过程有显著的作用.在分析并建立4个基本的过程性能模型之间的关系之后,以生命周期模型的定义为基础,提出了基于GQM的度量模型构建方法、资源模型构建方法以及质量模型构建方法,由于各过程性能模型的信息和数据都来源于生命周期模型,因而可以有效保证其之间信息与数据的一致.     

ARIMA模型在农产品价格预测中的应用

利用农产品价格时间序列的当前值和过去值准确预报未来值,将有利于正确引导农产品流通和农业生产,实现农产品区域供求平衡,并为政府和农户提供结构调整的依据。针对农产品价格这一重要问题,以白菜月价格数据为例,构建非平稳时间序列ARIMA（p,d,q）模型并预测白菜未来的月价格。结果表明ARIMA（0,1,1）模型能很好地模拟并预测白菜月价格趋势,为农产品市场信息的准确预测提供重要方法。     

基于时间序列模型的黄金价格预测

由于黄金价格受到诸多经济及政治因素的影响,其生成过程十分复杂,因此研究黄金价格的动态演变过程具有极强的现实意义。而具体的预测要根据市场的过去和当前的需求,以及影响市场需求变化的因素之间的关系,利用一定的判断、技术和模型,对其价格波动变化及发展趋势进行分析和判断。利用时间序列相关理论,建立黄金价格的ARMA模型,发现其可以动态刻画黄金价格数据的生成过程并且较好地对黄金价格进行预测。     

混合型PSO-BP模型在原油期货价格预测中的应用

建立了基于混沌理论下混合型PSO-BP模型,并运用此模型对纽约商品交易市场的原油期货价格数据进行了预测,并将预测结果与BP神经网络的预测结果进行了对比。结果表明混沌理论下混合型PSO-BP模型比单纯的BP模型具有较高的拟合度以及预测精度。     

基于图神经网络和长短期记忆模型的房价预测方法

针对目前仅单独考虑价格序列中样本的趋势或仅考虑多个关联属性与价格间的函数关系,而不能更准确地进行房价预测的问题,构建了时空注意力图卷积长短期记忆模型AG-LSTM,包含局部特征提取模块、区域特征提取模块、复合预测模块。局部特征提取模块分别使用同构图和异构图神经网络提取各小区及价格关系属性、各小区和配套邻居节点相关性的特征信息;区域特征提取模块先对邻近小区节点进行聚类,再结合图注意力网络获得小区节点对所属区域的重要性程度,建立区域与小区之间的映射矩阵,根据小区节点信息和映射矩阵得到区域特征;复合预测模块使用长短期记忆模型对由局部特征和区域特征组成的复合特征进行时序建模,实现房价预测。以链家网北京房价数据进行了实验,结果表明AG-LSTM预测结果优于已有基线模型。该模型同时挖掘了小区间位置关系、小区与其配套间位置关系、多个关联属性、价格时序趋势对房屋价格的影响,较好地实现了房屋价格的预测。     

资本市场系统结构模型应用研究

根据资产价格与交易者的相互影响机制,提出了资本市场结构模型及其数学原理,用于分析一组市场资产的价格演变过程,并对其拟合能力进行了探讨,随后进行了实证研究．理论研究和实证分析均表明,资本市场结构模型具有较强的整体拟合能力,能较好地对资产价格进行多指标预测．     

Optimal advertising decisions for promoting retail and online channels in a dynamic framework

Advertising plays an important role in sales promotion in supply chains. In this paper, advertising decisions for retail and online channels respectively, by retailer and manufacturer will be analyzed. The competition between retailer and manufacturer will be modeled by differential game theory, and feedback Nash equilibrium of the game will be obtained. Some insights are provided by using comparative statics. The obtained results show that compatibility factor of a product with online marketing, difference between manufacturer's online price and wholesale price, effectiveness of advertising, marginal profits, and cost of advertising all have significant effects on the advertising decisions in the equilibrium.     

Co-op advertising and pricing models in manufacturer–retailer supply chains

Cooperative (co-op) advertising plays a significant role in marketing programs in conventional supply chains and makes up the majority of promotional budgets in many product lines for both manufacturers and retailers. Nevertheless, most studies to date on co-op advertising have only assumed that the market demand is only influenced by the advertising level but not in any way by the retail price. That is why our work is concerned with co-op advertising and pricing strategies in distribution channels consisting of a manufacturer and a retailer. Four different models are discussed which are based on three non-cooperative games (i.e., Nash, Stackelberg retailer and Stackelberg manufacturer) and one cooperative game. We identify optimal co-op advertising and pricing strategies for both firms mostly analytically but we have to resort to numerical simulations in one case. Comparisons are then made about various outcomes, especially the profits, for all cases. This leads to consider more specifically the cooperation case in which profits are the highest for both the retailer and the manufacturer, and how they should share the extra joint profit achieved by moving to cooperation. We solve this bargain problem using the Nash bargaining model.     

Multi-product multi-market model for co-ordination of marketing and production decisions

A multi-product multi-market segment optimal control model for coordinating marketing and production decisions in a firm is presented. On the marketing side, the model considers pricing and the rate of advertising expenditure for each product for each market segment as decision (i.e. control) variables. On the production side, the production rate for an individual product is considered as a decision variable. The model assumes that demand for a product in each market segment is subject to seasonal and other exogenous effects. The model includes the following two special cases: (a) single-product multiple-market segments, (b) multiple products having purely exogenous, time-varying demand. The production planning subproblem in the model represents a class of partially singular optimal control problems. A solution procedure using Pontryagin's maximum principle is designed to solve the partially singular control problem. A hierarchical approach is then used in designing the solution procedure for the overall model.     

Developing a marketing decision model using a knowledge-based system

This paper describes using a knowledge-based system for developing a marketing decision model. The approach used in this study uses a decision table as a knowledge engineering tool. The decision table is used as a means of representing a set of decision rules to construct a developed marketing decision model. To support the modeling process, Prologa, an existing decision table engineering workbench, is used. The developed marketing decision model is used to determine the entrance time of a new product into market by utilizing knowledge-based systems. Presentation of a new product to the market at the best time will provide an advantage to competing companies and will increase their market share.     

Coordination of cooperative advertising models in a one-manufacturer two-retailer supply chain system

This paper considers cooperative advertising issues of a monopolistic manufacturer with competing duopolistic retailers. Four possible game structures (or power configurations), i.e., Stackelberg–Cournot, Stackelberg–Collusion, Nash–Cournot and Nash–Collusion, are discussed. Under each of four game structures, we develop a decision model for the three partners to design the optimal cooperative advertising policies. Through a comparison among the four models, we reveal how cooperative advertising policies and profits of all participants are affected by various competitive behaviors, and then determine whether the partners have any incentives to transit to a different structure. Also presented in the paper are a centralized decision model and a proposed cost-sharing contract, which is able to achieve perfect coordination of the considered channel, where the utility of risk preference is used to determine the fraction of local advertising costs shared by the manufacturer.     

Fuzzy analytical hierarchy process and multi-segment goal programming applied to new product segmented under price strategy

New product development (NPD) is becoming an important competitive advantage in the marketing strategies of current businesses. Developing a new product will incur fixed and variable costs, which then determine the product prices. Although this is a fundamental issue of marketing theory and practice, only a few papers on marketing models deal with price levels. The objective of this paper is proposing a model based on fuzzy analytical hierarchy process (AHP) and multi-segment goal programming (MSGP) to help decision makers to select the best pricing strategy for NPD. A case study of NPD under market selection strategy in multiple segment pricing levels for a Taiwan-based Watch Company is presented to illustrate the proposed methodology. The proposed method will guide the product development team to select the best market strategy by taking into account the price level and product/market segmentation.     

Marketing segmentation using support vector clustering

Marketing segmentation is widely used for targeting a smaller market and is useful for decision makers to reach all customers effectively with one basic marketing mix. Although several clustering algorithms have been proposed to deal with marketing segmentation problems, a soundly method seems to be limited. In this paper, support vector clustering (SVC) is used for marketing segmentation. A case study of a drink company is used to demonstrate the proposed method and compared with the k-means and the self-organizing feature map (SOFM) methods. On the basis of the numerical results, we can conclude that SVC outperforms the other methods in marketing segmentation.     

Selection of the optimum promotion mix by integrating a fuzzy linguistic decision model with genetic algorithms

Integrated marketing communication (IMC) is an important process by which a company can influence a target market, improve the position of that company’s product/service in the target market, and effectively build up its brand image. Sales promotion is an important communication channel designed to influence the customer’s purchasing behavior in the target market. There are many promotion tools available. Variations in business objectives and budgetary limits make it impossible for a company to employ all these promotion tools to convey sales messages to the customers. The selection of the best mix of promotion tools involves subjective information processing, instead of a numerically expressed objective decision-making process. In this research, we integrate a fuzzy linguistic decision model with a genetic algorithm (GA) to extract the optimum promotion mix of a variety of tools under satisfying expected marketing performance and budget limitations. The proposed methodology shows satisfactory results in an empirical study in terms of estimating the degree of satisfaction for achieving the business objectives, determining the optimum promotion mix, and minimizing expenditure on sales promotion activities.     

基于价格歧视的双寡头竞争企业网络广告投放模式测评与策略研究

双寡头竞争企业网络广告投放决策问题是网络营销企业面对的重要决策问题.鉴于此,重点考虑大众广告和定向广告两类网络广告,首先针对广告成本和消费者特征,给出用户群的效用函数和产品空间位置;然后,分别针对客户均匀分布和非均匀分布状况下的两类广告进行测评;最后基于测评结果深入研究定向广告的投放概率和策略选择问题.研究结果显示:无论消费者呈现均匀分布还是非均匀分布,定向广告的投入会使企业获利,但大众广告会同时降低两家竞争企业的利润,使得企业陷入囚徒困境,此外,实施价格歧视总是优于非歧视定价;网络广告的边际投放强度起到双向调节作用,定向广告的投放不总是有利的;当定向广告成本过高时,企业应当采用防御型策略;歧视定价时发放定向广告的数目总是高于非歧视定价.     

Coordinating static and dynamic supply chains with advertising through two-part tariffs

Zaccour (2008) analyses a marketing channel where firms invest in advertising to increase brand equity, showing that an exogenous two-part tariff cannot replicate the vertically integrated performance. I revisit the model proving that a multiplicity of efficient franchising contracts exists. I characterise an optimal two-part tariff specified as a linear function of the upstream firm’s advertising effort, performing this task both in the static and in the dynamic games. An analogous result emerges both in the static game, writing the fixed component of the two-part tariff as a non-linear function of the manufacturer’s advertising effort, and in the dynamic game, using a contract which is linear in the brand equity.