Mining status-set Sequential Pattern based on frequent itemset for failure prediction in a temporal database with multiple status items monitored

In this study, we investigate the problem of status sequential pattern mining (SSPM) based on frequent status set for failure prediction. We present a general sequential pattern mining framework with new definitions (e.g., frequent status itemset) and redefinitions (e.g., Sequence, Sequential Pattern) on sequential patterns for the field of failure prediction with multiple status items monitored. Some new indexes such as coverage rate (CR), hold rate (HR), and factor set (FS) are introduced to discover interesting Strong SSP and related factor set of some important status itemsets. The Apriori-like algorithms are also developed particularly for SSPM with high computational efficiency, and numeric examples are provided to demonstrate the process of SSPM for failure prediction. It shows that the proposed algorithm for SSPM is effective, capable of discovering meaningful sequential patterns with user-interested coverage rate and hold rate.