基於DBSCAN及Cox存活分析模型進行零售業顧客流失預測;Customer Churn Prediction applying to Retailer using DBSCAN and Cox Proportional Hazards Model

NCUIR > School of Management at National Central University > Graduate Institute of Business Administration > Electronic Thesis & Dissertation > Item 987654321/83377

Please use this identifier to cite or link to this item: http://ir.lib.ncu.edu.tw/handle/987654321/83377

Title:	基於DBSCAN及Cox存活分析模型進行零售業顧客流失預測;Customer Churn Prediction applying to Retailer using DBSCAN and Cox Proportional Hazards Model
Authors:	林家鋒;Lin, Chia-Feng
Contributors:	企業管理學系
Keywords:	資料探勘;存活分析;DBSCAN;Cox比例風險模型;DBSCAN;Cox Proportional Hazards Model;Data Mining
Date:	2020-07-17
Issue Date:	2020-09-02 15:32:53 (UTC+8)
Publisher:	國立中央大學
Abstract:	隨著顧客關係管理(Customer Relationship Management)的興起，越來越多的業者希望透過資訊科技來瞭解並滿足客戶的需求，建立與現有顧客更緊密的關係來提升顧客滿意度、降低顧客流失率，以提升企業的財務績效。因此，企業開始嘗試以資料探勘來進行對顧客進行分析，譬如使用分群演算法將顧客分成不同群體進行行銷，利用 BG/NBD 模型來預測顧客終生價值等。現今，顧客流失預測模型為 CRM 系統中常見的模型之一，可告知管理者哪些顧客將會離開企業，管理者即可針對此客群設計行銷活動等激勵措施，盡可能地降低顧客實際的流失率。此模型已是企業管理顧客流失的基本方法之一，現在常用於建立顧客模型的工具為羅吉斯迴歸(Logistic Regression)、類神經網路 (Neural Networks)、決策樹(Decision Tree)、支援向量機(Support Vector Machine, SVM)等。本篇論文提供零售業的顧客流失預測模型，透過 DBSCAN 演算法將現有客群的顧客交易紀錄依交易天數間隔進行分段，並以醫學存活分析研究中 Cox 比例風險模型建立顧客流失預測模型，可依據顧客前六十天的資料，預測顧客未來是否會流失。此模型可有效標記出可能流失的顧客，業者可針對這些顧客研究相關的行銷策略。;With the growth of CRM, companies have started to meet customer needs through information technology, and establish a closer relationship with existing customers to reduce customer churn rate, in order to improve the company′s financial performance. Therefore, companies began to use data mining to analyze customers, such as using clustering to divide customers into different cluster for marketing, and using BG/NBD models to predict customer lifetime value. Customer churn problem was originally the main topic from telecom industry. The customer churn prediction model is the solution which can provide managers which customers will be lost, and design the campaign for these customers, and this model is already the basic solution for companies to manage customer churn. The tools commonly used to build customer models are Logistic Regression, Neural Networks, Decision Tree, and Support Vector Machines. (Support Vector Machine, SVM), etc. In this paper, we provides a new method to segment customer transaction data of existing customer groups through the DBSCAN algorithm, and establish a churn prediction model based on the Cox proportional hazard model in medical survival analysis research, based on the previous 60 days of the customer Data to predict whether customers will be lost in the future.
Appears in Collections:	[Graduate Institute of Business Administration] Electronic Thesis & Dissertation

Files in This Item:

File	Description	Size	Format
index.html		0Kb	HTML	287	View/Open

社群 sharing

Loading...