【摘 要】
:
In the field of financial market behavior research, forecasting the change of stock market trend with classifier is very important.To improve the accuracy of stock market trend forecast, most recent r
【机 构】
:
Management School of Xiamen University Xiamen University
论文部分内容阅读
In the field of financial market behavior research, forecasting the change of stock market trend with classifier is very important.To improve the accuracy of stock market trend forecast, most recent research pays more attention to improve the parameter of classifier instead of digging out the character of input data.Stock market time series come from the reciprocity of multi-scale nonlinear factors which have essential multi-scale characteristics.Our work proposes a stock market trend forecasting method which deals with multi-scale time series features.The method is based on wavelet multiresolution analysis in order to carry out multi-scale decomposition of stock market time series data, and then our method extracts the memory and trend characteristics of multi-scale time series out.Finally, we use classifier to forecast stock market trend to improve accuracy of stock market trend forecast.
其他文献
近年来微信抢红包持续受欢迎,而这背后是微信抢红包用户强烈的参与动机,但是现有研究缺乏对于微信抢红包行为影响因素的实证分析.基于自我决定理论,本文构建了微信用户抢红包行为的理论模型,研究微信用户抢红包行为意向的影响因素.通过网上调研平台问卷星在微信群中收集了95份有效问卷,采用结构方程模型检验研究假设.实证研究结果表明,抢红包的自主感显著正向影响用户的持续参与意向,胜任感有显著负向影响,归属感的正向
With the "Internet plus finance" boom, the two industries have achieved a new level of convergence.Based on the perspective of two-way convergence, this paper firstly constructs the convergence framew
本文探讨建立以零售平台为主导者、零售商为跟随者的Stackelberg博弈模型.在平台和零售商联合促销的过程中,将平台对零售商的促销努力成本承担策略、价格折扣承担策略与不承担促销成本策略三者进行比较,分析不同的成本承担策略对零售商和平台的影响.研究结果表明,平台的成本承担策略提高了平台和零售商的利润和销量,证明其对供应链有协调作用,但不同策略的协调效果和适用范围不同,最后进行数值分析,证明了成本承
社会化商务中顾客购买或转换品牌很大程度上取决于口碑.对品牌商来说,用户发布的负面口碑可对品牌造成负面影响,品牌转换现象更严重.但负面口碑对品牌转换意愿的影响机制尚未得到系统分析.基于社会认知理论,研究探讨品牌认知和品牌情感在负面口碑对品牌转换意愿影响中的作用以更好地理解品牌转换意愿的产生机理,证实了品牌认知和品牌情感在负面口碑与品牌转换意愿之间的中介作用,可帮助品牌商降低顾客转换意愿,提高品牌忠诚
Existing studies on behavior prediction ignore peoples unobservable behavioral intents on location categories which determine their observable behavior destination and trajectories.Besides, environmen
In this work, we present a novel approach to estimate the daily commuting behaviors of mobile phone users.We integrate the signaling data of mobile phone with the information from the other resources,
近年来,随着电子商务的发展,网络营销模式拓展了安岳柠檬的销售渠道.但是从目前的销售数据显示,柠檬的网络销售量比重量占用还很小,生鲜果品缺少网络营销标准化体系.消费者缺乏对网购柠檬鲜果信心的关键因素是消费者认为的从电商到自己手中这个过程的未知性,既不能保证柠檬鲜果到达自己手中完好无损,也不明确种植过程中的农药残留情况.基于以上背景因素,本文从柠檬质量安全出发,针对柠檬供应过程质量问题,提出以柠檬产品
In the Internet+era, big data marketing, as a new application and new business model, subverts the competition form, changes the world rules for competition, and sets off the revolution in the whole i
在传统的基于内容的推荐算法中,用户模型和物品模型的构建都是以新闻、社交、购物、视频、图书等单一领域内的数据为基础,而对用户在交错领域内行为的综合分析尝试较少.推荐系统需要在大规模用户情况下依然具备快速产生推荐结果的能力,串行算法已经无法完成这个任务.本文使用中文信息学会社会媒体处理专业委员会提供的SMP2015微博数据集,在Spark集群环境下并行计算微博用户的TF-IDF特征向量作为跨领域用户模
近年来基于用户画像的精准营销越来越受工业界重视.为提高精准营销的质量,学术界也提出了多种提升用户画像生成的准确度的方法,而基于网络社群海量语料库进行用户画像——常住地理位置获取的方法也越发受到重视.这篇文章展现了一种基于海量微博用户语料对用户地理位置进行预测的新方法,不仅可对用户语料进行直接分析,还可大大提高预测结果的准确性.首先,通过研究发现,Boosting算法确实具有将弱分类器组合形成强分类