Research on fuzzy rules extraction of futures trading based on MapReduce

Due to the limitation of memory, time complexity and data complexity, common fuzzy rule extraction algorithms can only handle small or medium data sets. Parallel computing is one of the effective tools to deal with big data. This paper mainly proposes a parallel processing method to present how to extract fuzzy if-then rules from futures trading data in the programming model MapReduce. First, the proposed method combines K-means fusion FCM clustering algorithm to generate reasonable fuzzy sets based on the processed futures data. Then, the fuzzy rules with antecedent and consequent are extracted based on these fuzzy sets. Finally, by calculating the support degree of each rule, we can obtain the meaningful fuzzy rules used for futures trading. In the experimental studies, the feasibility and validity of the extracted meaningful fuzzy rules are verified by predicting the futures price trend.