Hardware Reference
In-Depth Information
4 Process Mining
Process mining has proven to be a valuable approach that provides new and objective
insights into the way processes are actually carried out within organizations [7]. Tak-
ing a set of real executions (a so-called "event-log") as a starting point, these tech-
niques attempt to extract non-trivial and useful information about the "real" process.
The central object in process mining is a particular operational process, such as a re-
view or change control process in a software development organization. Control over
these processes is often supported by information systems that help to coordinate the
steps that need to be performed in the course of the process. Examples of these infor-
mation systems are Document and/or Version Management Systems.
The process mining of the CCB process as a whole (i.e. 6870 cases) revealed that in
most of the cases (70%)
Analysis
task is skipped and the cases are being directly re-
solved [4]. We assumed that people tend to make such shortcuts in order to save time.
As the project progresses, people feel time pressure because of the approaching deadline
Our hypothesis is then that they decide to skip the
Analysis
task in order to save some
time. We expected that the number of the cases skipping the
Analysis
increases towards
the end of the project. In order to prove this hypothesis, we analyzed each lifecycle
phase separately. Table 1 shows the number of cases per each lifecycle phase.
Table 1.
Number of cases per lifecycle phase
Number
of
Number of cases
skipping the Analysis
cases
Specification
543
408
75.14 %
Design
477
368
77.15 %
Implementation
1282
998
77.85 %
Component testing
470
371
78.94 %
Integration testing
862
531
61.60 %
System testing
1759
861
48.95 %
Customer testing
81
48
59.26 %
Consumer use
33
20
60.61 %
Not Applicable
1363
1173
86.06 %
Although, we expected the number of cases that are not handled according to the
documented process increases towards the end of the project, the results of our analy-
sis do not prove that. The number of such cases is similar from the Specification
phase till the
Component testing
. A significant decrease is observed during the
System
testing
.
Customer testing
and
Consumer use
contain both too little cases for any con-
clusions to be made.
5 Task Duration
In our case study, we focused on the time aspect of the CCB process. Namely, we
analyzed the throughput time of the process and the duration of tasks per lifecycle
phase. We compared the duration of the three tasks
Analysis, Resolution
and
Evalua-
tion
. These tasks are described in detail by their
Start
and
End
events directly in the
