Customer segmentation system based on improved k-means and neural network clustering

Abstract

The invention discloses a customer segmentation system based on improved k-means and neural network clustering. The customer segmentation system comprises a bank client data collection module, a sample data extraction module, a clustering processing module and a customer class segmentation module; the bank customer data collection module is used for collecting bank customer data and storing the bank customer data in a bank network database; the sample data extraction module is used for randomly extracting samples from bank customer data in the bank network database and extracting 1 / 3 of data as sample data; the clustering processing module is used for adopting an improved k-means clustering method to perform clustering on each sample of the sample data and outputting a clustering result; the neural network training module is used for taking the clustering result as a training sample, adopting the neural network to calculate a weight of each layer of each attribute and obtaining a trained neural network; and the customer class segmentation module is used for inputting bank customer data into the trained neural network and performing segmentation on the bank customer. The customer segmentation system reduces a probability of extracting an isolate point in the sample, improves clustering accuracy and improves customer segmentation accuracy.

Description

technical field [0001] The invention relates to the field of data mining, in particular to a customer segmentation system based on improved k-means and neural network clustering. Background technique [0002] At present, the difference in value brought to banks by different types of customers is very obvious. By identifying and distinguishing such differences, banks can guide them to allocate market sales, service and management resources more reasonably, and obtain more with less investment. revenue, solving this problem requires customer segmentation. Bank customer segmentation means that banks classify customers according to their attributes, behaviors, needs, preferences and values ​​in a clear strategy, business model and specific market, and provide targeted products, services and marketing models. process. [0003] In related technologies, there are empirical classification methods and statistical analysis methods for bank customer segmentation. Bank customer segme...

AI Technical Summary

Problems solved by technology

However, the existing K-means clustering method cannot effectively avoid the contingency brought by a single random sampling method, the clustering stability is low, and it has fatal shortcomings of being sensitive to outliers

Images