Decision-making involves several processes such as data pre-processing, data reduction and data selection. In order to assure a valuable solution is made, each of these processes needs to be successfully conducted. When dealing with complex data, parameter reduction is one of the essential processes that the decision-makers should take into account. It helps to reduce the processing time, computational memory and data dimensionality in the decision-making process. However, some of the parameter reduction methods were unable to generate a sub-optimal value during the parameter reduction process. This problem could affect the performance of the classification process. Soft set theory is one of the parameter reduction methods that faces this kind of problem. As a result of the study, to enhance the capability of soft set parameter reduction method, an integration between soft set and rough set theories as a parameter reduction method had been proposed. It was based on the efficiency of these two theories in processing complex and uncertain data problems. These two methods were sequentially applied to simplify the initial parameters in order to improve the performance of the classification process. The experimental work had returned positive classification results and successfully assisted the standard soft set parameter reduction method in generating sub-optimal reduction set and also the classifier in the classification process.