奖励式众筹融资绩效的影响因素及预测研究

发布时间：2018-01-17 23:35

本文关键词：奖励式众筹融资绩效的影响因素及预测研究　出处：《吉林财经大学》2017年硕士论文　论文类型：学位论文

【摘要】：2009年,第一个奖励式众筹平台kickstarter在美国成立。在此之后,奖励式众筹平台迅速在全球范围内发展,在欧洲和美洲逐渐走向成熟并发展到了亚洲、中南美洲和非洲等各个地区。2011年7月,我们国家第一个奖励式众筹平台“点名时间”成立,这象征着我国众筹平台方面的开始。在接下来的几年里,很多不同模式的众筹平台迅速成立。据鸣金网数据中心不完全统计,截止到2015年11月底,被鸣金网收录的众筹网站有265个,其中奖励式众筹网站105个,约占39.6%。在项目的量上,奖励式众筹的数目最多。因此,本文利用众筹网(http://www.zhongchou.com)中的3468个奖励式众筹项目对影响其融资绩效的因素做分析,并且对奖励式众筹项目融资绩效进行预测研究,为我国奖励式众筹的发展提供理论和实证基础。本文采用众筹网平台中的交易数据对奖励式众筹融资绩效的影响因素做分析,并预测奖励式众筹的融资绩效。通过最优尺度模型分析奖励式众筹的显著影响因素和非显著影响因素,探索这些影响因素之所以显著和非显著的原因。再运用SOM算法将融资绩效离散化,目的是运用C5.0决策树算法、支持向量机算法和TAN贝叶斯网络算法对融资绩效进行预测分析。对比分析C5.0决策树算法、支持向量机算法和TAN贝叶斯网络这三种预测方法,得出C5.0决策树的预测最优。研究结果表明:项目名称长度、筹到的金额、项目所属的行业、项目所在的城市和评论数对融资绩效的影响较大。然而,筹资天数、发起人信息、是否有视频、关注数、支持数、项目更新、项目回报方式和项目回报时间是对筹资绩效的影响较小。C5.0决策树算法、支持向量机算法和TAN贝叶斯网络算法,这三种算法中,C5.0决策树的预测效果最好。另外,对于除了1类绩效以外的预测,C5.0的预测结果比支持向量机和TAN贝叶斯网络的预测准确度要高。基于理论和实际的研究,对于项目筹资人,本文建议筹资项目名称命名要合理和有吸引力,目标金额要根据实际情况来设置,筹资人要及时回复投资人的评论和建议,多增加互动。对于项目投资人,要注重中后续项目的发展以及奖励式众筹回报的方式。对于奖励式众筹行业,建议相关部门的监管需要进一步明确思路,该行业也要逐步走向规范。奖励式众筹平台需要探索新的模式,奖励式众筹可以扩展到一些新的行业。
[Abstract]:In 2009, the first reward crowdfunding platform kickstarter was founded in the United States. Since then, the reward crowdfunding platform has developed rapidly around the world. In Europe and America gradually matured and developed to Asia, Central and South America and Africa and other regions. July 2011, our country's first incentive crowdfunding platform "roll call time" was established. This symbolizes the beginning of our crowdfunding platform. In the next few years, many different models of crowdfunding platform were quickly established. There are 265 crowdfunding websites included by Ming Jin. Among them, 105 are reward crowdfunding websites, which account for 39.6. in the number of projects, the number of award crowdfunding is the largest. Therefore. This paper uses 3468 incentive crowdfunding projects in http: / / www.zhongchou.comto analyze the factors that affect its financing performance. And the incentive crowdfunding project financing performance prediction research. To provide theoretical and empirical basis for the development of incentive crowdfunding. This paper uses the transaction data in crowdfunding platform to analyze the factors affecting the performance of incentive crowdfunding. And forecast the financing performance of incentive crowdfunding. Through the optimal scale model to analyze the significant and non-significant factors of incentive crowdfunding. To explore the reasons why these factors are significant and non-significant, and then use the SOM algorithm to discretize the financing performance, the purpose is to use C5.0 decision tree algorithm. Support vector machine (SVM) algorithm and TAN Bayesian network algorithm are used to predict the financing performance. Three forecasting methods, C5.0 decision tree algorithm, support vector machine algorithm and TAN Bayesian network, are compared and analyzed. The results show that the length of the project name, the amount of money raised, the industry the project belongs to, the city where the project is located and the number of comments have great influence on the financing performance. Funding days, sponsor information, whether there is video, attention, support, project update, project return mode and project return time are the smaller impact on funding performance. C5.0 decision tree algorithm. Support vector machine algorithm and TAN Bayesian network algorithm, these three algorithms, C5.0 decision tree prediction effect is the best. In addition, for the prediction of performance other than class 1. The prediction accuracy of C5.0 is higher than that of support vector machine and TAN Bayesian network. This paper suggests that the name of the fund-raising project should be reasonable and attractive, the target amount should be set according to the actual situation, the fundraiser should respond to the comments and suggestions of the investors in time, and increase the interaction. We should pay attention to the development of the follow-up projects and the way of reward crowdfunding. For the reward crowdfunding industry, it is suggested that the supervision of the relevant departments need to further clarify the thinking. The industry also needs to be standardized. The reward crowdfunding platform needs to explore new models, and the reward crowdfunding can be extended to some new industries.
【学位授予单位】：吉林财经大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：F724.6;F832.39

【参考文献】