Analytical clustering procedures in massive failure data

The increasing accessibility of information through advanced communications and telecommunications devices have led to availability of massive data with large number of variables. The acquired information are then clustered or categorized based on specified objectives. The demand for variable clustering as a fundamental strategy for data analysis is increasing. Although a wide variety of variable clustering methods have been proposed, there is still need for advancements in this area. Many of the recommanded procedures have given only a narrow theoretical motivation, and their operational properties need more systematic investigation before they can be used with confidence. The problem of variable selection in large-scale datasets arises when it is desired to model the relationship between a response variable and a subset of explanatory variables, but there is uncertainty about which subset to use. In addition, in many areas, there are great interests in time and causes of events. Time-to-event data analysis is a kernel of failure analysis and has an inevitable role in predicting the probability of events occurrence. Thus, in the presence of massive and complex data, proper methods are required for efficient simplification of such data. This study is motivated by the importance of the applied variable clustering in complex high-dimensional failure data to facilitate data analysis. Analytical procedures using combinational heuristic algorithms for variable clustering are proposed. The purpose of these procedures is to cluster the explanatory variables and identify sets of most correlated variables in such datasets.