In this paper, the map-reduce technique is utilized for speeding up the mining process and derived as similar results as our previous approach. The chromosome representation consists of four parts that are a mapper number, grouping part, stock part and portfolio part. According to mapper number, chromosomes in population are divided into subsets and sent to respective mappers. Fitness evaluation and genetic operations are the same with our previous approach, and executed on reducers. The evolution process is repeated until reaching the terminal conditions. Experiments are conducted on a real dataset to show the performance of proposed approach.
the ASE BigData & SocialInformatics 2015, Article 41