Optimization of Overdraft Application Process with Fluxicon Disco

Having reliable information is very important while using a process mining technique as data mining is all about finding out how processes are actually performed to discover problems and areas for improvement. In this paper, we analyzed the real-life event log from a financial institution using the data mining tool called Fluxicon disco. The dataset is related to the application process for a personal loan or overdraft. The event log has a total of 262200 events and 13087 cases. The data was collected from October 1, 2011, to March 14, 2012. The whole loan procedure started with the customer submitting an application and the successful endpoint is the application being activated. For this study, we collected the data in CSV format and imported it into the Fluxicon Disco software. The Disco makes it easier to deal with large event logs and complex models. After importing the data, we defined the column as Case ID, Activity, Resource, Timestamp, and Attributes. We do not have any large specific problem; the dataset simply shows the loan process activities and path. But we assume that there might be some departments and activities which might be taking longer than necessary, and it might be increasing the entire loan selection process. We assume that some work is not being done in an efficient way, resources of the organization are not optimized, and as a result, clients have to wait for more for the loan process. So based on these problems, our objective is to find those activities and paths where duration is comparatively very high, and cases are being piled up there. But the data analytics tool does not exactly state what the real problem is, it just gives us direction and we have to find it out on our own. In this data analyzing process, we came up with two major areas which we think need to be taken care of. The first one is the activity which is modifying approved contracts where the frequency is very low, but the mean duration is very high. And it is placed in a bit of a weird place which is after application and offer events are completed. So, we assume that it is the result of faults in the above process and things can be improved in the application and offer selection process either. Another finding is the path from W_Nabellenoffertes-COMPLETE to W_Nabellenoffertes-START (which basically means following up with client start and complete) is taking a very high total duration and a lot of cases are being piled up here. So, we assume that this particular path and activity needs to be looked up to find if the work is being done in an efficient way.