Page 42 - 2023-Vol19-Issue2
P. 42
38 | Andreswari, Millenia, Rizky, Haniyah & Mufti
because it is only used to estimate the relationship between tribute. Each Case ID has several ongoing activities such
two activities or events so that they can find out how closely as Reassignment, Update from customer, Assignment, and
related one event class is to another event class. Of the three others.
matrices (unary significance, binary significance, and binary
correlation), the first will provide information to simplify Timestamp is an attribute that marks the time the activity
the initial model and will be used to build process models occurs in each case. Each activity has its own timestamp.
adaptively (Gu¨nther & van der Aalst, 2007). Timestamp is very important to determine the order in which
an activity occurs from start to end of a Case ID. The times-
2) Adaptive Graph Simplification Process: In the graph tamp attribute is indicated by Datetamp which has a time
simplification process with Fuzzy Miner, there are three pro- value in the form of the date and time the activity occurred.
cesses: binary conflict resolution, edge filtering, and node
aggregation and abstraction. The binary conflict resolution Originator, is an attribute that indicates the actor or person
process is influenced by 2 parameters, namely the preserve in charge of each activity that takes place. In the event log that
threshold and the ratio threshold. The edge filtering process is used as the originator is the Assignment Group attribute
is influenced by 2 parameters, namely the edge cutoff and which contains the name of the team as the perpetrator of each
utility ratio. Meanwhile, the node aggregation and abstraction activity (Table II).
process is influenced by the cutoff node parameter [26].
TABLE II.
Fuzzy miner is also implemented to handle heterogeneous FINAL EVENT LOG DATA
logistics learning processes. Process mining is able to extract
the learning path into the process model from beginning to Case ID Timestamp Activity Originator
end and can present process variants. Because there is an IM0000004 07/01/2013 08:17 Reassignment TEAM0001
outlier based on the number of occurrences when a group of IM0000004 04/11/2013 13:41 Reassignment TEAM0002
students interact with the learning outcomes, the outlier data IM0000004 04/11/2013 13:41 Update from customer TEAM0002
must be removed first to get the appropriate results [27]. IM0000004 04/11/2013 12:09 Operator Update TEAM0003
IM0000004 04/11/2013 12:09 Assignment TEAM0003
IM0000004 04/11/2013 13:41 Assignment TEAM0002
III. RESULTS AND DISCUSSION A. Discovery
Discovery data is done with Disco to get insightful event log
A Preprocessing is done by cleaning data such as noise
reduction, removing duplication of data, removing unused data (Fig. 2). The steps taken in the discovery process include
attributes, and data transformation. There are 466,737 data importing data, setting parameter attributes, displaying the
with seven initial attributes that are displayed in the event process model, and exporting the results of the discovery in
log in CSV format, namely Incident ID, Datetestamp, Inci- MXML format. The process model generated at the discovery
dentActivity Number, IncidentActivity Type, KM number, stage is called the initial process model, which is obtained
Assignment Group, and Interaction ID (Table I). In this pro- based on a matrix with cases transformed into activity nodes.
cess, 4 main attributes are used to be processed for modeling, Initial process models.
namely Case ID, Activity, Timestamp, and Originator.
TABLE I.
INITIAL EVENT LOG DATA
Incident ID DateStamp IncidentActivity Number IncidentActivity Type Assignment Group
IM0000004 07/01/2013 08:17 001A3689763 Reassignment TEAM0001
IM0000004 04/11/2013 13:41 001A5852941 Reassignment TEAM0002
IM0000004 04/11/2013 13:41 001A5852943 TEAM0002
IM0000004 04/11/2013 12:09 001A5849980 Update from customer TEAM0003
IM0000004 04/11/2013 12:09 001A5849979 Operator Update TEAM0003
IM0000004 04/11/2013 13:41 001A5852942 Assignment TEAM0002
Assignment
Case ID is an identity that determines the scope of the Fig. 2. Initial process model
process. In this study, the scope of the process is the activity
in the incident section, so that the Incident ID column is used The initial result process model is in the form of spaghetti
as Case ID. Each Case ID represents a process that is different processes caused by too many relationships between activities.
from other processes. Case ID will be simplified into numeric
form for easier reading. Activity, is an attribute with a value
that describes the stages of the running process. The attribute
that represents the activity is the IncidentActivity Type at-