一种加权聚类划分决策树算法

针对斜划分决策树算法普遍存在时间效率低、部分算法仅能应用于二分类问题，提出了一种基于加权距离的聚类决策树算法。通过Relief-F算法为预测属性计算权重，并将权重用于树结点中数据的聚类过程，使用分簇结果对结点进行多路划分，得到可直接用于多分类问题的决策树。理论分析和实验结果表明，该算法与经典轴平行决策树相比，拥有更好的泛化能力以及相近的算法时间复杂度，与大部分斜决策树相比，在付出更少计算代价的前提下，获得了近似的正确率以及模型简洁度。     

Does the spatial representation affect criteria weights in environmental decision-making? Insights from a behavioral experiment

The need and interest to consider cognitive and motivational biases has been recognized in different disciplines (e.g. economics, decision theory, risk analysis) and has recently reached environmental decision-making. Within this domain, the intrinsic presence of a spatial dimension of both alternatives and criteria calls for the use of maps throughout the decision-making process to properly represent the spatial distribution of the features under analysis. This makes spatial Multi Criteria Decision Analysis (MCDA) a particularly interesting domain to explore new dimensions of cognitive biases. This study proposes a behavioral experiment aimed at discovering to what extent the spatial visualization (i.e. maps) of criteria versus the non-spatial one (i.e. tables) can bias the weight elicitation phase of a spatial MCDA process. The experiment simulates a very common analysis in environmental and land use planning: land suitability analysis. Our findings show that there are significant consequences on how important we perceive a certain criterion to be, depending on whether it is represented as a map or as a table among a mix of maps and tables. Indeed, the map representation of the same criterion leads to higher weights attributed to that criterion compared to the table representation. Visualizing the same information as a map or as a table, although technically equivalent, is thus not psychologically equivalent for Decision Makers. The results of this experiment are expected to have implications for spatial decision-making processes, by generating better awareness on the impacts of map-mediated land suitability analysis.     

Disentangling the drivers of renewable energy investments: The role of behavioral factors

The scope of this study is to investigate the main determinants of renewable energy investments in Greece. For this reason, we build an appropriate framework to infer the decision‐making process in the renewable energy sources (RES). The main drivers are then categorized under two groups as economic, and behavioral ones. In the next stage, we rely on binomial (logit and probit) and quantile regression analysis to estimate the impact of these factors on investment decisions towards RES. Our findings indicate that investors who gain better access to knowledge and exhibit a more positive attitude towards the contribution of artificial intelligence (AI) on the RES industry have a higher share of renewables in their portfolio. We argue that these investors are willing to invest a higher amount in the RES industry today. At the same time, quantile regression models illustrate that this relationship is nonmonotonic (i.e., inverted U shaped curve), arguing that positive attitudes towards the contribution of behavioral factors on RES are not adequate to encourage investments after crossing a certain point (threshold). Based on the empirical findings, we argue that the message to policymakers and government officials is to promote a faster penetration of low carbon technologies to achieve sustainable economic and social growth.     

基于改进TOPSIS法的供应商选择模型研究

为有效地帮助企业快速找到合适的供应商合作伙伴,采用直觉模糊集、评分函数等方法对TOPSIS评估法进行优化,并以此为基础建立了一种供应商选择模型。首先,采集和评估供应商的产品质量、产品价格、产品交货的可靠性、供应位置、财务情况、库存水平、劳资关系、发展能力和技术能力等相关信息,由专家给出主观评估信息,汇总为综合属性值;然后,通过直觉模糊熵确定各评估指标的权重;最后,综合考虑供应商选择决策过程中的多个目标和标准,应用改进TOPSIS法的对供应商进行分类选择。结果显示,基于改进TOPSIS法的供应商选择模型能够较准确地反映出各供应商的真实水平和对企业的潜在价值,可以有效地解决不确定条件下对供应商的选择问题,提高了供应商选择结果的可靠性。改进后的模型简便易行,具有良好的稳定性,对于合理制定企业供应商选择标准以及进一步优化决策模型具有一定的借鉴意义。     

基于Stacking模型融合的店铺销量预测

为了实现企业产品销量预估,提高生产供应的准确性与效率,提出了基于Stacking模型的融合算法进行销量预测。算法设计了两层堆叠的模型结构,初级学习器采用随机森林、支持向量回归、差分整合移动平均自回归、轻量级梯度提升机器和门控循环单元5种单模型,将分类与回归树作为次级学习器构成Stacking融合模型,并对数据进行了预测。预测结果显示,使用Stacking模型融合后得到了较好的预测结果,比单模型中效果最好的模型的均方根误差更小,平均绝对误差更小,决定系数值更大,表明Stacking融合模型的预测准确率更高。所设计模型可用于对企业店铺的产品销量进行预测,帮助企业更好地安排生产、营销活动,为减少库存、缩短生产销售周期提供数据支持,对企业生产决策有一定的参考价值。     

抢抓成渝“双圈”机遇打造川南渝西联动发展新支点的思考

根据中央成渝双城经济圈建设的战略决策和四川省、泸州市相关部署,结合天华公司“十四五”规划和现实情况,认真学习贯彻落实上级部署要求和讲话精神,学以致用,融会贯通,抢抓自身优势,科学谋划,主动融入双城经济圈建设,积极推进公司高质量发展,实现天华的二次腾飞和再塑辉煌。     

The Role of the Global Reporting Initiative's Sustainability Reporting Guidelines in the Social Screening of Investments

Social screening of investments calls not only for investment policy and criteria, but also for information about companies, their policies, practices and performance. The Global Reporting Initiative (GRI) and its June 2000 Sustainability Reporting Guidelines have the potential to significantly improve the usefulness and quality of information reported by companies about their environmental, social and economic impacts and performance. The GRI aims to develop a voluntary reporting framework that will elevate sustainability reporting practices to a level equivalent to that of financial reporting in rigour, comparability, auditability and general acceptance. This will be a welcome and efficient supplement to the questionnaires, interviews, press releases, media reports and other sources of information traditionally used for screening in investment decision making – social/ethical and mainstream. The Dow Jones Sustainability Group Index, the Jantzi Social Index and the Innovest EcoValue'21 analytical platform, together with the SRI community, are all likely to benefit from GRI-style sustainability reports. One of the GRI's key challenges is to accommodate the broad variety of disclosure needs and expectations of a wide range of report users and company stakeholders. To maximize the usefulness of the GRI Guidelines, report users, including the SRI community, need to be engaged in the process of developing and refining the Guidelines over time. The GRI Guidelines are emerging as an important instrument in enabling companies to communicate with their stakeholders about performance and accountability beyond just the financial bottom line.     

Does Greater Amount of Information Always Bolster Attitudinal Resistance?

Previous research suggests that attitudinal resistance to information that challenges a prior evaluation increases with the amount of information underlying the prior evaluation. We revisit this proposition in a context in which a set of important claims about a target brand are presented either alone—a lower amount of isolated information—or along with other favorable, but less important claims—a higher amount of embedded information. Results from two experiments show that when the challenge occurs immediately after the initial evaluation, a greater amount of embedded initial information does produce greater attitudinal resistance. However, when the challenge occurs after a delay, a lesser amount of isolated information produces greater attitudinal resistance. The findings qualify previous assumptions about the role of prior information in attitudinal resistance, and support a constructive view of judgment revision and attitudinal resistance.     

论投资者决策行为模式中的倾向效应

行为金融研究发现 ,投资者的决策行为并非完全理性 ,“倾向效应”普遍存在于个人投资者和投资专业人士中。了解“倾向效应”及其心理动因具有积极的现实意义     

An Analysis of the Factor Structure of Jones’ Moral Intensity Construct

In 1991, Jones developed an issue-contingent model of ethical decision making in which moral intensity is posited to affect the four stages of Rest’s 1986 model (awareness, judgment, intention, and behavior). Jones claimed that moral intensity, which is “the extent of issue-related moral imperative in a situation” (p. 372), consists of six characteristics: magnitude of consequences (MC), social consensus (SC), probability of effect (PE), temporal immediacy (TI), proximity (PX), and concentration of effect (CE). This article reports the findings of two studies that analyzed the factor structure of moral intensity, operationalized by a 12-item Perceived Moral Intensity Scale (PMIS) adapted from the work of Singhapakdi et al. [1996, Journal of Business Research, 36, 245–255] and Frey [2000, Journal of Business Ethics, 26, 181–195]. The two items that were purported to measure CE were dropped due to their inability to effectively tap into the characteristic proposed by Jones. Factor analyses of the remaining 10 items supported a 3-factor structure, with the MC, PE, and TI items loading on the first factor, the PX items loading on the second factor, and the SC items loading on the third factor. These factors were labeled: Probable Magnitude of Consequences, Proximity, and Social Consensus. The authors conclude that moral intensity consists of three characteristics, rather than the six posited by Jones.