运用移动端数据采集与挖掘的渠道管控研究

发布时间：2018-01-15 00:15

本文关键词：运用移动端数据采集与挖掘的渠道管控研究　出处：《浙江理工大学》2017年硕士论文　论文类型：学位论文

【摘要】：渠道是制造商将产品发送到中间商、零售商最终到达消费者手中的通道或途径,是产品或服务从产地转移到其它地方整个过程所依赖的组织结构。由此可见:(1)渠道管控高度依赖渠道成员;(2)渠道管控高度依赖渠道终端的营销数据。本文涉及的中烟A公司全国零售户超过50万,销区覆盖全国。该公司为实现终端数据移动采集、确保资源投放准确、提高销售决策响应度,通过渠道管控移动应用的开发,实现渠道有效管控,并构建渠道风险预警管控机制。本论文研究目标:通过对渠道管控移动端应用的开发,满足企业对渠道信息迅速获取的需求,并运用移动终端数据采集功能达到渠道零售户数据快速回收的目的;以数据挖掘为手段,科学解决数据采集过程中地域广有重叠、巡点多难分配、实地采集工作量估算难、采集品种目标锁定难的问题;运用数据挖掘结果,提出有效的渠道管控方法。依据研究目标,完成的研究任务如下:(1)完成对渠道管控移动应用的开发,实现通过基于ESB的Web Service架构服务平台的数据请求,实现用户通过对数据项的下钻、维度选取等方式的渠道主题信息数据获取。(2)将移动终端零售户数据采集巡店计划中的店铺选定,通过K-means聚类对店铺坐标进行区域圈定。确定人员对区域一对N关系,将人员采集区域收敛、分类;同时提出改进的k-medoids算法,应证并优化聚类结果减少重叠率,加强收敛。(3)运用因子分析法对多维空间的数据进行降维处理,简化采集的数据。在销区零售户内、在千余种快消产品中,锁定渠道主要管控对象和每日数据采集目标;并对区域内零售商类型进行聚类分析。(4)对移动终端零售户数据采集功能增加属性收集层,收集终端渠道成员(巡店业务员)巡点数、巡点覆盖面积、时间三个维度属性,用于定量分析;并以123位巡店业务员的工作属性为例,运用fisher分类算法用训练数据创建的模型,对新增若干名业务员按其工作属性判别归类。(5)基于数据挖掘结果,结合中烟A公司实例,提出渠道成员相应的管控决策。包括:ABC分类、80/20规则、平滑指数法等管控方法。
[Abstract]:A channel is a channel or channel through which a manufacturer sends a product to an intermediary and a retailer finally reaches a consumer. It is the organizational structure on which the whole process of product or service transfer from place of production to other places depends. Thus, channel control is highly dependent on channel members; The marketing data of channel management and control are highly dependent on channel terminal. In this paper, the retail households of Zhongyan A Company are more than 500,000, and the sales area covers the whole country. The company is to realize the mobile collection of terminal data. To ensure accurate resource delivery, improve the sales decision-making response, through the development of channel management and control mobile applications, achieve effective channel control. This paper aims: through the development of channel management and control mobile applications to meet the needs of enterprises to obtain channel information quickly. And use the mobile terminal data collection function to achieve the channel retail household data fast recovery purpose; By means of data mining, the problems of overlapping regions, difficult distribution of patrol points, difficult to estimate the workload of field collection and difficulty to lock the target of collecting varieties are solved scientifically in the process of data acquisition. Using the results of data mining, an effective channel control method is proposed. According to the research objectives, the research tasks are as follows: 1) the development of channel control mobile application is completed. The data request is realized through the Web Service architecture service platform based on ESB, and the data item is drilled by the user. Dimension selection and other ways of channel topic information data acquisition. 2) the mobile terminal retail household data acquisition tour shop plan in the shop selection. K-means clustering is used to delineate the location of the shop coordinates. A pair of N relationship between the personnel and the region is determined, and the area of personnel collection is converged and classified. At the same time, an improved k-medoids algorithm is proposed, which should prove and optimize the clustering results to reduce the overlap rate, and strengthen the convergence. 3) using factor analysis to reduce the dimension of the data in multidimensional space. Simplify the data collected. In the retail area, in the thousands of fast elimination products, lock the main control channels and daily data acquisition targets; Cluster analysis of regional retailers. 4) add attribute collection layer to mobile terminal retail household data collection function, collect terminal channel members (shop operators) patrol points, patrol coverage area. Three dimensional attributes of time, which are used for quantitative analysis; Taking the working attributes of 123 tour salespeople as an example, the model created with training data using fisher classification algorithm is used. Based on the results of data mining and the example of Zhongyan A Company, this paper puts forward the corresponding management and control decisions of channel members, including the classification of ": ABC". 80/20 rule, smooth index method and other control methods.
【学位授予单位】：浙江理工大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：F721;TP311.13

【相似文献】