Predicting factors affecting likelihoods in engineering problems

A method of predicting the factors which most affect the likelihoods of specified events from data bases or tables of data is presented. This method is called the Variable Precision Rough Sets Model with Asymmetric Bounds. The data is broken into a limited number of meaningful ranges. The tables of data are then reduced to a minimum set of variables and rules for predicting the likelihood of the specified event. The method has been developed in a mathematically precise form. The methodology determines the best variables and data patterns to categorize the data in terms of the likelihood of a given event. This procedure has been programmed and tested on a steel industry problem. However, the procedure is applicable to any problem where an outcome or event is to be predicted from a set of known variables provided the data is available in a tabular or data base form. Such problems would include any failure or reliability problem in power, control or electronic systems. Another problem to which this method could be applied is that of predicting which input variables most affect the output variables in a complex control system.