Page 40 - 2023-Vol19-Issue2

36 |
to very small details of log activity [11],[12],[13]. Process
mining plays a role in modeling and analyzing processes
by finding, monitoring, and improving the actual process
by extracting knowledge from available event logs. Process
mining can define the model process as well as the relationship
between the actual process and the data. There are three stages
of the mining process, namely discovery, conformance, and
enhancement [14].
Discovery plays a role in analyzing event log data by creat-
ing an initial process model without using a priori information.
Conformance checking is used to check whether the model
created matches the actual event. Enhancement extends or
augments the process model with real-life information. The
enhancement process can be carried out in two forms, namely
repair or extension. Repair is done by modifying the model
to better reflect reality, while the extension is done by adding
a new perspective based on the available event logs and the
resulting process model [5].
The discovery process is done with Disco tools while
conformance with ProM. Disco is also used to sort out unused
data, remove noise, and filter out invalid information to the
endpoint [15]. ProM is used to implement the algorithm on the
model generated by Disco [16]. The algorithm used in process
mining for the analysis of Rabobank Group ICT incidents is
the Fuzzy Miner Algorithm and the Heuristic Miner. Event
log data on Rabobank Group ICT incidents is complex real-
life company data that requires a special approach to handling
it. Fuzzy Miner and Heuristic Miner algorithms can be used
in process mining to be able to handle complex event logs so
that they can be observed properly [17],[18]. Disco and ProM
are tools that can be used in data processing with statistical
capabilities, filtering functions, and process map generators
on Disco as well as innovative process mining capabilities on
ProM [19],[20],[21].
This study aims to provide an analysis of the Rabobank
Group ICT incident based on the 2014 BPI Challenge event
log with process mining using the Fuzzy Miner Algorithm and
Heuristic Miner through the Disco and ProM tools to predict
the workload for the service desk so that improvements can be
made to the implementation of software releases as Rabobank
Group ICT business development efforts.
In a previous study ”Analysis and Implementation of Pro-
cess Mining with Heuristic Miner Algorithm Case study:
Event logs of Rabobank Group ICT Netherlands” an inci-
dent analysis was carried out on the event log of Rabobank
Group ICT with the Heuristic Miner algorithm using Filter
Log using Simple Heuristics on ProM. Based on the test re-
sults, it is known that the ideal process model is obtained by
setting the positive observation threshold parameter of 1000,
relative to the best 0.05, and the dependency threshold value
of 0.9. The parameter that has the greatest influence is the pos- for evaluation by Rabobank Group ICT (Fig. 1).
Andreswari, Millenia, Rizky, Haniyah & Mufti
itive observation threshold. This is because these parameters
have the ability to filter data with very small trace frequencies
so that they are not processed in the mining process [3].
Another study ”Analysis and Implementation of Process
Mining Using Fuzzy Mining (Case Study: Data BPI Chal-
lenge 2014)” which showed that the use of a threshold in the
simplification process resulted in good conformance with a
low preserve threshold, a high utilization ratio. high, and
the edge cutoff and node cutoff are low. Based on the test,
the process model with the best conformance is a preserve
threshold of 0.05, utility ratio of 0.85, edge cutoff of 0.05, and
node cutoff of 0.05 [10].
Based on the results of previous studies, it is known that
the review of process mining results only focuses on measure-
ment parameters and does not provide process models regard-
ing incidents that occur. For this reason, this study provides a
process model from the parameters that have been determined
in previous studies to find out the problems that occur in the
Rabobank Group ICT incident from all aspects of the event
log that affect changes in the workload of the service desk/IT
operations. In addition, additional tools, namely Disco, are
used to generate valid data up to the endpoint by filtering and
reducing the noise before entering ProM. In addition, process
modeling in ProM with Heuristic Miner is carried out with
interactive Data-aware Heuristic Miner (iDHM) with the best
parameter settings for event logs automatically (Mannhardt
et al., 2017). This setting will be compared with the results
of manual trials conducted [3] with a Filter Log using Simple
Heuristics on ProM.
II. METHODS
The decline in the number of product marketing is a serious
problem and an obstacle to increasing the number of software
releases at Rabobank Group ICT. This affects the business
development of Rabobank Group ICT. For this reason, it is
necessary to analyze to overcome the incident that occurred.
The Service Desk Agent (SDA) records the activities that
occur in the form of an event log to be analyzed and given
solutions to overcome the incidents that occur. Process Min-
ing is one way that can be used. The complex data makes
the Fuzzy Miner and Heuristic Miner algorithms chosen to
provide a process model for incidents that occur. Thus, im-
provements can be made to the IT service desk/operations to
help Rabobank Group ICT return to a steady state.
The process mining system was built by creating an initial
process model at Disco and implementing the Fuzzy Miner
algorithm with Mine for a Fuzzy Model and the Heuristic
Miner algorithm with interactive Data-aware Heuristic Miner
(iDHM) on ProM on the Rabobank Group ICT incident log
activity. The output built is a process model that can be used