摘要
随着经济的快速发展，企业竞争日益激烈，合适的店铺选址是影响企业发展的重要因素之一，它能为企业后期发展奠定坚实的基础。由于连锁店具有分享知名度、口碑等优势，商家在竞争市场中会优先选择经营连锁店。目前与连锁店选址相关的文献数量较少并且提出的选址研究方法中存在许多待完善的内容。因此，本文将研究对象细化到火锅连锁店，对如何制定出一种更完善，更准确的火锅连锁店选址模型进行了探讨。
本文基于数据挖掘对火锅连锁店选址模型进行了研究。在选址评价指标体系的建立过程中，首先提出了初始评价指标体系，它由5个一级指标和15个二级指标构成。一级指标分为人口因素、交通因素、竞争因素、自身因素和其他因素。一级指标细分时使用最小圆覆盖算法、经纬度距离公式、聚类分析和Kruskal-Wallis秩和检验。然后，本文使用caret包和boruta包中的特征选取算法，对初始评价指标体系内的特征进行选取。特征选取结果为，caret包保留11个特征，boruta包保留12个特征，从而更新选址的评价指标体系。确定了三种不同的选址评价指标体系后基于武汉地区数据以及南京地区数据构建并验证火锅连锁店选址模型。本文的研究结论为：（1）基于三种评价指标体系分别使用logistic 回归、神经网络和支持向量机三种算法建模，将建模结果汇总后进行对比得到准确率最高的模型，该模型的准确率为92.5%，召回率为75.8%，模型使用的算法是支持向量机，其中核函数为多项式核函数。对应的评价指标体系是使用boruta包特征选取得到的，由5个一级指标和12个二级指标构成。（2）确定火锅连锁店选
址模型后，使用南京地区数据进一步验证模型性能。模型验证结果显示，火锅连锁店选址模型的准确率为90.3%，说明本文提出的火锅连锁店选址模型准确率高，具有实用性。
本文创新点在于完善和丰富了连锁店选址的评价指标体系，添加了自身因素和竞争因素，基于优化后的选址评价指标体系构建火锅连锁店选址模型，模型准确率高，适用范围广。本文的研究方法和研究结果扩展了选址研究的思路，具有实际应用的价值。
关键词：连锁店选址；logistic回归；神经网络；支持向量机
Abstract
With the rapid development of economy and increasingly fierce competition among enterprises, suitable store location is one of the most important factors affecting the development of enterprises, which can lay a solid foundation for the later development. Because chain stores have the advantages of sharing popularity and reputation, businesses will be preferred to operate chain stores. At present, there are few literatures related to the location of chain stores and there are many contents which need to be improved in the proposed research methods. Therefore, this thesis refines the research object to hotpot chain stores and discusses how to develop a more perfect and more accurate hotpot chain store location model.
Based on data mining, this thesis studies the location model. In the process of establishing the evaluation index system, the initial evaluation index system is firstly proposed, which is composed of 5 first-level indexes and 15 second-level indexes. The first-level indicators are divided into population factor, traffic factor, competitive factor, self factor and other factors. The minimum circle coverage algorithm, latitude and longitude distance formula, clustering analysis and kruskal-wallis rank sum test are used in the subdivision of the first-level indexes. Then, this thesis uses the feature selection algorithm in caret package and boruta package to select the features in initial evaluation index system. The results of feature selection are that caret package retains 11 features and boruta package retains 12 features. Three different evaluation index systems are determined, and the location model of hotpot chain stores is constructed and verified based on wuhan regional data and nanjing regional data. Research conclusions of this thesis are: (1) Logistic regression, neural network and support vector machine are used for modeling based on the three evaluation index systems. After summarizing the modeling results, the model with the highest accuracy is obtained. The accuracy of the model is 92.5% and the recall rate is 75.8%. The algorithm used in the model is support vector machine, and the kernel function is polynomial kernel function. The evaluation index system consists of 5 first-level indicators and 12 second-level indicators. (2) After the hotpot chain location model is
determined, the performance of the model is further verified with the data of nanjing region. The results of model validation show that the accuracy of hotpot chain location model is 90.3%, indicating that the hotpot chain location model proposed in this thesis has high accuracy and practicability.
The innovation of this thesis lies in improving and enriching the evaluation index system of chain store location, adding self factor and competitive factor, and building the hotpot chain location model based on the optimized evaluation index system. The model has high accuracy and wide application range. The research method and results of this thesis extend the idea of site selection research and have practical application value.
Key words：Chain Store Location Selection; Logistic Regression; Neural Network;
Support Vector Machine
目录
摘要 ................................................................................................................... I Abstract ............................................................................................................... I I 第1章绪论 (1)
1.1 研究背景 (1)
李川 娄艺潇表达豪情壮志的诗句1.2 研究目的与意义 (2)
1.2.1 研究目的 (2)
1.2.2 研究意义 (2)
1.3 国内外研究现状 (3)
1.3.1 选址的研究现状 (3)
1.3.2 分类算法的研究现状 (5)
1.4 研究内容与研究方法 (6)
霸气的网络游戏名字1.4.1 研究内容 (6)
1.4.2 研究方法和技术路线 (7)
第2章火锅连锁店选址的指标选择 (10)
2.1 研究区域和研究对象的确定 (10)
2.2 影响火锅连锁店选址的主要因素分析 (11)
2.3 火锅连锁店选址指标的细分 (12)
2.3.1 人口因素细分 (12)
2.3.2 交通因素细分 (14)
2.3.3 竞争因素细分 (14)
2.3.4 自身因素细分 (22)
2.3.5 其他因素细分 (25)
第3章选址评价指标体系的特征选取 (26)
3.1 特征选取的概念 (26)
3.2 选址的初始评价指标体系 (27)
3.3 基于caret包的特征选取 (28)
3.4 基于boruta包的特征选取 (30)
第4章火锅连锁店选址模型的构建 (33)
4.1 算法细节的确定 (33)
4.1.1 算法介绍 (33)
4.1.2 神经网络隐层数、隐层节点数和激活函数的选择 (35)
4.1.3 支持向量机核函数的选择 (36)
4.2 数据预处理 (37)
4.3 模型准确性评价体系 (38)
4.3.1 准确率与召回率 (38)
4.3.2 ROC曲线与AUC值 (39)
4.4火锅连锁店选址模型的性能分析 (39)
孙涛老婆个人资料简介
4.4.1 logistic回归的建模结果 (39)
火锅连锁4.4.2神经网络的建模结果 (41)
4.4.3支持向量机的建模结果 (43)
4.4.4 模型性能对比分析 (44)
第5章模型验证 (46)
5.1 模型验证的数据准备 (46)
5.1.1 模型验证地区选择的依据 (46)
5.1.2 连锁店数据的采集 (47)
5.1.3 实验数据的整理 (48)
5.2 模型的验证过程 (50)春望原文
5.3 实验结果分析及应用策略 (51)
5.3.1 实验结果分析 (51)
5.3.2 应用策略 (53)
第6章总结与展望 (54)
6.1 全文总结 (54)
6.2 研究成果与创新点 (55)
6.3 研究展望 (55)
致谢 (57)
参考文献 (58)
攻读硕士学位期间发表的学术论文 (63)
附录A (64)