Linguistic summaries of sales data at a computer retailer via fuzzy logic and a genetic algorithm

We present an implementation (as an add-in to the DBMS) of fuzzy linguistic summaries of sales data of a small-to-medium size computer retailer in Southern Poland. The system derives linguistic summaries exemplified by “about a half of sales of accessories is in autumn” or “about 1/3 of sales of network elements is in the beginning of the year”. An inherent imprecision of natural language is accounted for by using elements of fuzzy logic, and to effectively and efficiently search the company´s database, a genetic algorithm is employed. The fitting function represents a multiaspect measure of the quality of a linguistic summary. The results obtained turned out to be useful to the owner, and suggested some changes in sales policy, determining the staff size, negotiation of commissions, etc. A combination of fuzzy logic and a genetic algorithm proved to be effective, efficient and synergistic