基于多源大数据融合的银行网点选址方法

针对传统银行网点选址方法中存在的人为主观因素较大、数据量支撑不够、考虑因素理想化等问题，提出一种基于多源大数据融合的银行网点选址方法。该方法通过多源数据构造人流量、交通拥堵指数、用户价值、周边竞争网点数和人均收入5个基础特征，并利用协同训练的半监督学习方法扩充训练集。基于基础特征与机器学习算法构建多个子模型，将子模型的输出概率作为特征，构建基于逻辑回归的集成算法，作为银行网点选址模型，同时提出一种优化银行网点权重的损失函数，以保证模型预测中更佳的银行网点具有更高的权重。通过实验分析表明，该算法相较于传统算法预测评估更为准确，能够很好地解决银行网点选址问题。

Site selection method of banking facility location based on multi-source big data fusion

In order to solve the problems in the traditional method of bank facility location, such as subjective human factors, a single source of data, and idealized factors, this paper proposes a bank location selection method based on multi-source big data fusion. The method constructs five characteristics by multi-source data, namely, human traffic,traffic congestion index,user value, surrounding competitive network points and per capita income, and uses semi-supervised learning method of collaborative training to expand the training set. Based on the basic features and machine learning algorithm to build a number of sub models, the output probability of the sub model is taken as the feature, and the integrated algorithm based on logical regression is constructed as the location model of bank outlets. At the same time, a loss function is proposed to optimize the weight of Bank outlets, so as to ensure that the better bank outlets in the model prediction have higher weights.The experimental analysis shows that the algorithm is more accurate than the traditional algorithm, and it can be well applied to the bank location problem.