Fuzzy mining of meteorological data

The process of knowledge discovery is essentially that of randomization. The key to discovery lies in the successful mining of non-trivial attributed databases. Randomizing declarative data for procedural knowledge (i.e., the theory) will necessarily be limited to the syntax of if-then rules. A logic for randomization underpins every theorem defined in any formal consistent system. A set-based heuristic logic for the randomization of a database is applied to a meteorological forecasting problem. The term fuzzy mining implies a heuristic process for database randomization, rather than an admissible process, which is usually intractable. The representation of knowledge in a database is seldom exact. The goal is to rapidly find rules of maximal generality (i.e., so as to maximize their predictive value where not all field values need be specified). Discovered rules may be translated into SQL. In this manner, the records in any database that satisfy the conditions (i.e., the independent variables) in the specified rules can be selected