Transforming Timestamp of a CSV Database to an Event Log

Process mining can help organizations easily capture information from enterprise transaction systems and provides detail. The very first step of process mining is availability of event log which can be obtained by preprocessing event log from other data or databases. Preprocessing event log divided into three parts, namely the transformation of the database to event log, filtering activities of event log, and separating the event log based on day, week, month, and year. The transformation is carried out to convert the data in the TOS (Terminal Operating System) database into an event log containing the process of importing goods at the Container Terminal in Surabaya, Indonesia. Transformation is carried out to obtain the contents of event log needed, such as process ID, activity or message, actor, execution time, cost, and activity or message supporting attributes. This research will focus on how to convert the timestamp in the TOS database into the timestamp of event log, namely single timestamp and double timestamp. The two main methods proposed in this paper are firstly, inverse of the normal cumulative distribution and secondly, by using sojourn time. The experimental results showed that using the proposed method can transform the timestamp of TOS database into a single timestamp and double timestamp of event log.