Abstract:

SelfOrganizing Map(SOM) is a competitive unsupervised learning neural network method. The SOM neural network has been widely used in pattern classification and recognition, and topology invariance mapping, etc. This paper delves into the application of SOM in the clustering analysis of China's 31 provinces and constructs an enrollment decision model for independent colleges. First, it selects registration rate, first wish rate and per capital GDP in each province as the  input mode of SOM; then, it uses SOM to cluster; finally, it obtains the features and level of student source by analyzing the clustering results. The experimental results show that SOM is feasible and effective to the cluster analysis of student sources; and it can avoid subjective factors and more rapidly and  objectively get the clustering results. It provides a new reference basis for making a list of enrollment plan and an enrollment propaganda program for independent colleges, which has potential prospects in the field of enrollment for independent colleges.

Key words: neural network;SOM;Kohonen;enrollment