基于分位数回归的C2C电商卖家销量分析与信用评价

发布时间：2018-04-11 06:09

本文选题：C2C电子商务 + 分位数回归　；参考：《合肥工业大学》2017年硕士论文

【摘要】：近年来,C2C电子商务事业发展迅猛,以淘宝网为首的C2C电子商务网站的营业额每年以极快的速度提高。同时,互联网、移动网络在市场上的推进和虚拟支付手段的不断完善,也为C2C电子商务的发展提供了高速通道,越来越多的消费者选择在网络上完成交易。随着网络消费者群体的不断增加,C2C电商已经形成了一个新的研究领域,同时在其高速发展的过程中也产生了许多问题。鉴于此,已有研究针对C2C电商领域中的问题进行了挖掘与研究,其中电商中的销量问题和信用评估问题占据了很大一部分研究工作的工作内容,同时也取得了较好的研究结论与成果。然而,已有的文献对此的研究工作,主要使用均值回归分析方法,难以考虑C2C电商问题中在不同分位水平下有可能出现的异质性情况。为此,本文使用分位数回归方法,对C2C电商卖家的销量与信用开展研究,挖掘不同分位水平下的异质现象与变动规律,以期获得新的发现。本文主要开展了以下两个方面的实证研究工作。(1)对C2C电商店铺销量的影响因素进行了分析与探讨。使用了神经网络分位数回归方法,通过抓取淘宝网上的实际店铺交易数据,进行了实证研究。研究发现:最低价格、最高价格、卖家信用得分、累计评论数这几个因素对店铺销量会产生异质性影响,影响因素对于不同销量水平的网上店铺会产生不同程度甚至不同方向的影响。(2)对C2C电商卖家信用评分的方式进行了探讨并提出了有效性检验的方法。研究中通过分析现有卖家信用评分方式的不足,提出了卖家信用评分的多因素校正模型。进一步,为了检验模型的实际效果,提出了可以衡量C2C电商店铺对消费者造成损失的指标——店铺消费累计损失,并建立了以神经网络分位数回归模型为基础的卖家信用评分模型的有效性检验方法。通过抓取淘宝网实际销售数据进行实证研究,发现建立的多因素校正模型确实能够更加有效地反映电商店铺的真实信用水平。本文的研究工作,对C2C电商店铺销量影响因素的探讨,可为卖家制定营销策略提供理论依据与决策参考,有助于卖家针对性地开展营销工作,提高销量。对C2C电商卖家信用评分方式的改善,能够更加有效地进行卖家信用水平的评估,有利于卖家提升店铺信用水平。
[Abstract]:In recent years, C2C e-commerce business has been developing rapidly, and the turnover of C2C e-commerce websites, led by Taobao, is increasing at a very fast rate every year.At the same time, the promotion of the Internet, mobile network in the market and the continuous improvement of virtual payment means, also provide a high-speed channel for the development of C2C e-commerce, more and more consumers choose to complete transactions on the network.With the increasing number of online consumers, C2C e-commerce has formed a new research field, and many problems have arisen in the process of its rapid development.In view of this, some researches have been carried out in the field of C2C e-commerce, in which the sales volume and credit evaluation have occupied a large part of the research work.At the same time, good research conclusions and results have been obtained.However, the existing literatures mainly use the method of mean regression analysis, it is difficult to consider the heterogeneity of C2C ecoquotient problem at different quantiles.Therefore, this paper uses the quantile regression method to study the sales volume and credit of the sellers of C2C e-commerce, and excavates the heterogeneity and variation law under different quantile levels in order to obtain new discoveries.This paper mainly carried out the following two aspects of empirical research work: 1) the impact of C2C e-commerce store sales factors were analyzed and discussed.The neural network quantile regression method is used to analyze the real store transaction data on Taobao.The study found that the lowest price, the highest price, the seller's credit score and the cumulative number of comments had heterogeneous effects on store sales.The influence factors have different degree or even different direction to the online shop of different sales level.) the way of credit score of C2C e-commerce seller is discussed and the method of validity test is put forward.Based on the analysis of the deficiency of the existing seller's credit rating method, a multi-factor calibration model of seller's credit score is proposed.Furthermore, in order to test the actual effect of the model, this paper puts forward an index to measure the loss caused by C2C e-commerce stores to consumers-the cumulative loss of store consumption.The validity test method of seller credit rating model based on neural network quantile regression model is established.Through the empirical study of Taobao's actual sales data, it is found that the established multi-factor correction model can reflect the true credit level of e-commerce stores more effectively.The research work of this paper discusses the influencing factors of sales volume of C2C e-commerce stores, which can provide theoretical basis and decision reference for sellers to formulate marketing strategies, help sellers to carry out marketing work and increase sales volume.The improvement of the credit scoring method of C2C e-commerce seller can evaluate the seller's credit level more effectively, and it is helpful for the seller to improve the store credit level.
【学位授予单位】：合肥工业大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：F724.6;F274

【相似文献】