Stock networks community structure division algorithm based on multi-gene families encoding
Kangshun Li · Computer Engineering and Science · 2013
In order to solve the problems of the traditional stock networks community structure division , such as low searching accuracy , high time complexity and ease of going into a local optimal solution , this paper proposes a new community division algorithm based on Multi-Gene Families ( MGF ) encoding GEP to research the stock market complex networks community structure division phenomenon. The algorithm takes advantage of the MGF property to respectively encode the stock node ID and the community type into two different MGFs , and then implicitly encodes the relationship of the two MGF into the chromosome through a mapping function.Meanwhile , the elite migration strategy is applied to the whole hereditary stage e.g.gene selection , chromosome crossover , chromosome inversion , restricted permutation and so on , which can prevent premature and speed the convergence.Experimental analysis shows that this algorithm implements the stock complex network division accurately and efficiently , and the community structure division result can provide the stock investors with profound information.