An analysis of the rule weights and fuzzy reasoning methods for linguistic rule based classification systems applied to problems with highly imbalanced data sets