A novel cluster HAR-type model for forecasting realized volatility

This paper proposes a cluster HAR-type model that adopts the hierarchical clustering technique to form the cascade of heterogeneous volatility components. In contrast to the conventional HAR-type models, the proposed cluster models are based on the relevant lagged volatilities selected by the cluster group Lasso. Our simulation evidence suggests that the cluster group Lasso dominates other alternatives in terms of variable screening and that the cluster HAR serves as the top performer in forecasting the future realized volatility. The forecasting superiority of the cluster models are also demonstrated in an empirical application where the highest forecasting accuracy tends to be achieved by separating the jumps from the continuous sample path volatility process.     

A product-centric data mining algorithm for targeted promotions

Targeted promotions in retail are becoming increasingly popular, particularly in UK grocery retail sector, where competition is stiff and consumers remain price sensitive. Given this, a targeted promotion algorithm is proposed to enhance the effectiveness of promotions by retailers. The algorithm leverages a mathematical model for optimising items to target and fuzzy c-means clustering for finding the best customers to target. Tests using simulations with real life consumer scanner panel data from the UK grocery retailer sector show that the algorithm performs well in finding the best items and customers to target whilst eliminating “false positives” (targeting customers who do not buy a product) and reducing “false negatives” (not targeting customers who could buy a product). The algorithm also shows better performance when compared to a similar published framework, particularly in handling “false positives” and “false negatives”. The paper concludes by discussing managerial and research implications, and highlights applications of the model to other fields.     

浙江省民营企业集群式营销探讨

民营企业集群已成为浙江省区域经济快速发展的主流模式之一。与大型企业、跨国公司一起,已成为参与市场竞争的主角。因此,市场营销理论应用领域已从单个企业营销逐步延伸到民营企业集群式营销。民营企业集群式营销是民营企业集群成长的关键,浙江省民营企业集群的快速发展与其营销管理创新是分不开的。为了提高浙江省民营企业集群营销管理水平,本文着重探讨了浙江省民营企业集群营销战略、策略制定及组织实施问题。     

创新驱动下我国R&D研究演化及热点主题分析

无论是创新驱动战略实施还是科技创新能力提升都离不开R&D,通过R&D研究演化发现热点主题,可为后续研究提供参考借鉴。通过对我国1998-2018年以R&D为主题的文献进行梳理,借助文献计量学方法,运用信息可视化分析工具,对文献的关键词共现密度、关键词共现聚类和突变词等信息进行可视化图谱分析,研究R&D文献演化过程及热点主题。研究发现,20年来我国以R&D为主题的文献研究沿着“宏观面—微观面—宏观面”演化推进,并且政府补贴、融资约束、战略性新兴产业及创新效率与R&D的关系分析成为当前研究热点。     

集群式产业链形成与演化内在机理研究

论文在分析诱发因素的基础上,对集群式产业链形成的内在自增强机制——分工、集群与产业链互动关系进行剖析,指出这是一种具有双刃剑性质的可逆机制;论文还重点构建基于群链租金增进的动力机制与群链耗散的阻力机制模型,在此基础上提出并论证集群式产业链形成与演进内在机理的一般过程,并对群链生命周期演化各阶段的特点及其演化原因进行诠释。     

The spatial and temporal variation in passenger service rate and its impact on train dwell time: A time-series clustering approach using dynamic time warping

This study aims to improve the understanding of the underlying mechanism of passenger boarding and alighting processes, as well as its potential influence on train dwell time and train operation. Empirical data collected from one of busiest metro stations in Seoul, Korea, demonstrates the spatial and temporal variation in the passenger service rate, as a result of interference between boarding, alighting, and standing passengers. This study postulates that the level of interference can be associated with the train car occupancy and the proximity of train door to entry points, as waiting passengers tend to cluster near the platform entries. A unique temporal service rate curve is derived from each door location. We introduce Dynamic Time Warping for similarity assessment and clustering. It revealed four groups of train doors showing distinct shapes of curve from each platform. The first cluster includes the train doors located closest to the platform entry points where the initial service rate is drastically impeded by severe inference among passengers. The level of interference gradually diminishes as the door is located farther away from the entry points, but the overall service rate decreases as well. A passenger service time model is derived to include the cluster variable. To test its significance, the prediction capability of the model is presented with and without the cluster variable. The results of this study may be used to guide waiting passengers along the platform to minimize interference and to avoid serious disruption during passenger service time.     

过渡期产业群的演化与政策启示——以浙江永康五金产业群为例

当前我国产业集群正处于过渡时期,市场机制的完善提高了市场交易主体间扩展信任水平,从而降低基于扩展信任的交易费用水平。随着集群规模不断扩大,理性的企业会选择更多地与传统集群外有竞争力的企业进行交易,从导致产业集群由以血缘与亲缘等为纽带的传统集群向基于扩展信任的现代集群演进。本文以浙江永康五金产业群为例对此进行了分析,提出了产业集群发展的几点方略。     

基于内容相似性的个性化推荐系统的应用

推荐系统应用于远程教育平台中，有助于学生学习。推荐系统具有在线和离线推荐两部分，在线推荐使用最近邻居完成，离线推荐采用基于语义相似的推荐算法，利用信息内容和用户兴趣的相似性来分析数据仓库中相关数据针对用户兴趣模型找到相似文档进行推荐。     

To cluster or not to cluster?: Understanding geographic clustering by restaurant segment

This study tested whether geographic clustering differs by restaurant segment due to the differences in consumers’ hedonic and utilitarian values by using Ripley’s K function and a Tobit model. This study found that higher priced restaurant segments have stronger clusters than lower priced restaurants, which implies that restaurants that focus on hedonic values tend to cluster more than utilitarian focused restaurants. However, the results differ depending on whether or not restaurants are located within a central business district (CBD). For example, quick service restaurants have stronger clusters than casual restaurants outside CBDs. Practical applications may apply to new restaurants that are attempting to open. Up-scale restaurants have the advantage of reducing research costs by locating near similar restaurants. Moreover, casual restaurants do benefit by clustering near existing ones under the condition that demand is not severely hurt by competition, while quick service restaurants benefit by diffusing from each other.