Hybrid Parameterisation Model for Missing Datasets

Abstract

Missing datasets usually exist in many fields such as medical diagnosis, traffic controlling, meteorology, business, industrial process, computer and network telecommunication. This missing data might also decrease the efficiency of results during decision making process. Besides, missing data may lead to difficulties in making decisions. Therefore, an efficient method such as parameterisation is required to deal with these problems. Probability, heuristic, and machine learning are among the approaches that have been proposed in generating an optimised attribute set. However, some of the proposed works only consider certain problems to be solved and failed to analyse certain types of data. The aim of this study is to propose a hybrid parameterisation model that is capable to deal with missing datasets. Experimental results have shown that the proposed model is significant to be implemented in handling missing datasets. It also proves that processing time and memory space could be reduced while assisting the classifier in gaining high performance results.