A Project Manager Skill-up Simulator Towards Problem
Solving-based Learning
Masanori Akiyoshi
, Masaki Samejima
and Norihisa Komoda
Faculty of Applied Information Science, Hiroshima Institute of Technology, 2-1-1 Miyake Saeki-ku, Hiroshima, Japan
Graduate School of Information Science and Technology, Osaka University, 2-1 Yamadaoka, Suita, Osaka, Japan
Project Management, Skill-up Simulator, Scenario Generation, Association Rule Mining.
This paper addresses a project manager skill-up simulator which aims to provide problem-solving learning
environment. Project management is inherently human-centric activities, and research work for education has
been done by using simulation. The proposed simulator environment is designed to provide well-configured
functionalities that make it possible to generate high fidelity scenario on project situation with event rule firing
mechanisms, evaluation on a learner’s operations as to the system development progress management phase.
The proposed environment realizes effective scenario generation and well-insighted evaluation along with a
learner’s interactive operations.
Recently we have faced with the problem occurred
in the project management of IT system develop-
ment such as delivery delay and cost increase, even
if development organization and division are well-
prepared at the project planning phase. Project man-
agers are strongly expected to handle various situa-
tions at the project fields based on deep understand-
ing of “body of knowledge on project management”,
for instance, PMBOK(PMI, 2009) (Project Manage-
ment Body Of Knowledge) and so forth. As for the
consolidation of “body of knowledge on project man-
agement”, we have a world-wide used examination by
PMI (Project Management Institute) and could eval-
uate whether project managers have it or not. As for
handling various situation at the real project, OJT (On
the Job Training) is mostly applied to train such skills.
Success and failure factors of projects(Horine, 2005)
are also argued. However OJT has drawbacks that it
sometimes causes risks against practical projects and
depends on the ability of tutors as excellent project
This paper addresses a concrete methodology to
enhance problem solving ability of learners using our
developed project simulator. The simulator-based
learning environment have been proposed for project
management(Davidovitch et al., 2007)(Drappa and
Ludewig, 2000), for instance, “SESAM”(Drappa and
Ludewig, 2000) provides interactive functions to
manage the system development project. The simu-
lator just mimics some aspects of real projects and
the learners can make task assignment to members in
the simulator and receive evaluation of the final re-
sult. The learners may judge members’ skills through
Q&A (Question and Answer) style communication in
the simulator. Of course it is important to mimic real
project aspects in the simulator, besides we believe
more significant factors when using a simulator are to
provide intended scenario generation along with the
learning objectives set by tutors and detailed feedback
on each learner’s operations. “intended scenario gen-
eration” is not patterned one, but dynamically induced
one by using project model and a learner’s interactive
This paper proposes such a new type of project
manager skill-up simulator which realizes scenario
generation mechanism with event rule firing and
feedback mechanism concerning a learner’s defec-
tive operation identification from “reference opera-
tion” viewpoints.
Akiyoshi M., Samejima M. and Komoda N..
A Project Manager Skill-up Simulator Towards Problem Solving-based Learning.
DOI: 10.5220/0004140401900195
In Proceedings of the International Conference on Knowledge Management and Information Sharing (KMIS-2012), pages 190-195
ISBN: 978-989-8565-31-0
2012 SCITEPRESS (Science and Technology Publications, Lda.)
2.1 Specific Features on Problem
Solving-based Learning
Project management includes several phases, for in-
stance, planning, progress management, negotiation
to clients and so forth. Negotiation involves uncer-
tainty coming from human relationships, so we focus
on the phases common to planning and progress man-
agement hereafter. As for the target phases, required
abilities of project managers are logical thinking-
based actions to check usable resources and promised
constraints, and make appropriate operations towards
the situation. In a sense, PDCA (Plan-Do-Check-
Action) cycle is repeated until finalizing problem-
solving towards the situation.
Then what is the approach for providing such a
PDCA cycle as a scenario? Patterned scenario makes
it possible for tutors to set intended learning objec-
tives explicitly. On the other hand learners opera-
tions are so restricted to follow the patterned scenario,
which means the simulator just generate the same
results along with the intended learning objectives
and seems not to contribute to skill-up as “problem-
solving”. Unexpectedness is necessary for learners
to do the PDCA cycle with logical thinking, which
makes it difficult to guarantee the learning objectives
set in advance by tutors.
Based on the above-mentioned features on
“problem-solving in the project management”, the is-
sues to be addressed are as follows.
Learning scenarios are generated from the simula-
tor when tutors simply set the learning objectives.
Tutors get the reference operations as a solution
concerning the learning scenario even if they are
either optimal or semi-optimal.
Learners make similar operations which mimic
essential ones in real projects under recognizing
the learning objectives.
Learners get appropriate feedback as to their op-
erations based on evaluation of problem-solving
2.2 Configuration of PM Skill-up
In the section 2.1, the PDCA cycle is argued on “plan-
ning” and “progress management” as a prerequisite
for the simulator. Hereafter our proposed simulator
addresses the “progress management” from the is-
sues of “problem solving viewpoints, because the
model and related functionalities are more concrete
than “planning” ones. Figure 1 shows the configura-
tion of the project manager skill-up simulator.
Figure 1: Configuration of PM Skill-up Simulator.
The project model mainly consists of “Project”,
“Module”, and “Person” as they exist in the real
project. “Project” defines a set of modules to be de-
veloped and members to be assigned to the project.
The dependency of modules is depicted as “parent
module” and “child module”. “Module” defines each
estimated man-hour and technical domain such as
“database”, and its difficulty grade ranked with A
(difficult)”, “B (normal)”, and “C (easy)”. “Person”
defines each member’s skill grade on technical do-
mains ranked with A (expert)”, “B (normal)”, and
“C (novice)”. These three level description on module
difficulty and person skill generates project dynamics,
e.g. mismatching between module difficulty and per-
son skill is main cause of schedule delay and quality
loss in a project. And the most crucial task ofa learner
as a project manager is to detect/predict such unde-
sirable situation and make proper operations, for in-
stance, “overtime directive”, “collaborative work di-
rective with expert members” and “member assign-
ment change”. “collaborative work directive with ex-
pert members” is especially found in the real situation
of debugging. The process of debugging generates
new bugs and some low-skilled level members cannot
solve such iterative chain of generating bugs without
supervised collaboration. The simulator also models
latent bugs which makes it difficult to estimate final
product quality. The result is evaluated using indices
such as Q(uality), C(ost), and D(elivery), where “Q”
corresponds to the total bugs, “C” does the overtime
cost, and “D” does the completed date against the ini-
tially scheduled date.
What we explained so far is normal situations in a
sense, since the project model generates daily project
dynamics and operations by a learner are mostly nor-
mal reactions if he/she has knowledge on project man-
agement such as PMBOK and experiences on real
projects such as small delay and standard volume oc-
currence of bugs. In addition to imitated model of
a project, “event rules” are used to make some dis-
turbances towards project dynamics that reflect the
learning objective by tutors. For instance, the learn-
ing objective is “quality management is tough on
the project”, then unexpected explosive occurrence of
bugs is triggered by the corresponding event rule fir-
ing to the situation.
Scenario is generated based on the above-
mentioned intertwined interaction of model-driven
dynamics, event rule firing, and a learner’s operations.
Project status is displayed using “Gantt chart panel”
and operations are done through “interaction panel”
As shown in Figure 1, “Event rule set genera-
tor” and “Learner-oriented feedback are necessary
mechanisms of our proposed simulator to realize un-
patterned scenario-based learning. “Event rule set
generator” is to generate necessary event rule set from
all event rules so that the learning objective is guar-
anteed. “Learner-oriented feedback” is to detect a
learner’s defective operations by comparing them to
“reference operations” and provide related questions
to a learner.
2.3 Event Rule Set Generator
“Event rule” is defined as disturbance to cause un-
expected effects on the project status that reflect the
learning objectives set by a tutor. Figure 2 shows the
descriptive forms for depicting such a rule. For in-
stance, if the learning objective is concerning “Qual-
ity Management”, the simulator should generate the
situation on quality” even if a learner makes appro-
priate operations and vise versa. As indicated in Fig-
ure 2, descriptive parts are combinations of each part
items, that is, 27 rules are used. Of course all the
rules are not necessary to guarantee the project sta-
tus, which means the number of rule set is 2
. There
inherently exist so many variances as to the project
status in the simulator under intertwined interaction
of project model dynamics and a learners operation.
Therefore it can not be feasible to set such a rule set
by hand.
Figure 3 shows our proposed mechanism to in-
duce such a rule set by using agent programs that
perform various operations against the project sta-
tus. Agent programs are characterized to make either
Figure 2: Description of Event Rule.
Figure 3: Event Rule Set Generator.
proper or improper operations for the learning objec-
tive. First “Event rule selector” sets a certain set of
event rules. Then by using “project model”, “agent
programs”, and the “selected rule set”, the simula-
tor outputs the project status log and “QCD (Qual-
ity, Cost and Delivery)”. “Situation evaluator” judges
whether the intended project status is generated along
with the learning object which gives a score on the
“selected rule set”. Again a pair of “selected rule set”
and its score is inputted to the “Event rule selector”.
This repetitive process continues until one “rule set”
as optimum is decided. To realize this optimization,
GA(Genetic Algorithm) is used.
2.4 Learner-oriented Feedback
The project result from “QCD” viewpoints is output
of the simulator which is derived from a learner’s
operations. Simple evaluation of a learner is done
by using these indices, however, it does not lead to
skill-up from “problem solving” viewpoints. Excel-
lent project managers are considered to have judg-
ment criteria, so called “project management princi-
ple”, and make proper operations(hereafter, we call
them “reference operations”) along with it. There-
fore acquisition of such criteria is positioned as one
of the goals when providing the “problem solving-
based learning”. If a learner makes some mistakes
on his/her operations with lack of such criteria, the
learning simulator should point out the mistakes and
give related questions to confirm his/her knowledge
on it. The “reference operations” are derived from
applying the “project management principle” to the
project status in the simulator. So this paper assumes
that such “project management principle” and “refer-
ence operations” are known in advance, to focus on
the feedback problem to a learner.
2.4.1 Mechanism Overview
The feedback consists of two mechanisms, e.g. “iden-
tification mechanism on learner defective operations”
and “question generation mechanism”. Figure 4
shows the configuration of the feedback, using a
learner’s log and a reference operation-induced log.
The log includes several set of results with project
models, because “a learner’s defectiveness” is de-
tected across several project models.
Figure 4: Overview of Learner-oriented Feedback.
2.4.2 Difference Extraction
The difference of operations is done for each module
log in which the following items are used.
the times of “overtime directive” and “collabora-
tive work directive”
first operation timestamp
The predicates are “more/same/less” for the times
and “late/not late” for the timestamp.
The data representing the difference includes the
above items, module name, operation type, predi-
cates, and the project status attributes indicated in Ta-
ble 1.
2.4.3 Identification Mechanism
There exist three types of defective operations; 1) de-
fective ones frequently happened in several project
Table 1: Project Status Attributes.
Name Value
module difficulty three level; “A”, “B”, “C”
module work volume real number
skill mismatch logical value; true, false
delivery delay logical value; true, false
is on critical path logical value; true, false
models, 2) defective ones frequently happened in a
certain project model, and 3) defective ones happened
by chance. The defective operations frequently hap-
pened in several project models are the target opera-
tions which are caused by the lack of knowledge on
“project management principle”.
Figure 5 shows the process flow for identify-
ing our target operations. Association rule mining
process generates association rules between “project
status attributes” and “difference predicates”, which
means applying these rules makes it possible to fil-
ter “defective operations happened by chance”. Then
among the generated association rules, exact match-
ing ones to “project management principle” are ex-
tracted. The extracted rules makes it possible to filter
defective operations frequently happened in a certain
project model.
Figure 5: Process Flow of Identification.
2.4.4 Question Generation Mechanism
Questions related to defective operations frequently
happened in several project models are generated by
using templates with “choice list style”. The exam-
ple on output dialogue from the simulator is shown in
Appendix A.
3.1 Result of Event Rule Set Generation
To verify the intended scenario generation by our pro-
posed mechanism, we assume the followingcondition
to the simulator. A tutor sets “quality loss project” as
the learning objective, and 5 modules and 3 members
are used as the 30 days-long project model. Table
2 shows the generated event rules on the descriptive
Table 2: Generated Rules on Quality Loss.
Event Level
on Quality
Additional bugs M
on Cost
Work efficiency S
on Quality
Additional bugs L
As indicated in Table 2, the part of “Evaluation
function is used as to “Estimation on Quality” that
means disturbances from bugs management are trig-
gered when “quality” is well-maintained. Therefore
this will make such a situation to care about “quality”
and do some operations for it. We also evaluate how
many times of the “quality loss” project status hap-
pens in the project as to agent programs entirely. Av-
erage times of such project status is 2.67, and it seems
to be reasonable for the reason that “quality loss” sta-
tus happens every 10 days in the 30 days-long project.
3.2 Result of Learner-oriented
We used 10 project models for 2 learners; learner-A
and learner-B. Investigation by hand detected 3 de-
fective operations as to the learner-A and 2 defective
operations as to the leaarner-B. Then our proposed
mechanism extracted rules as shown in Table 3.
Table 3: Extracted Rules.
learner-A learner-B
initial association rules 99 38
rules for identification 2 1
Table 4 shows the identified rule and the num-
ber of generated questions. As indicated in Table 4,
unidentified defective operation still remains; “ade-
quate skill against the module collaborative work
directiveis many”. This was caused by the reason that
various action part descriptions for the condition part
“adequate skill against the module” existed in the log,
which causes filtering such association rules by our
proposed mechanism.
Table 4: Generated Questions.
learner rule(number of question)
low skill against the module
collaborative work directive is few (7)
low skill against the module
collaborative work directive is late (6)
low skill against the module
collaborative work directive is late (5)
This paper addresses a project manager skill-up sim-
ulator which aims to provide problem-solving learn-
ing environment. The distinctive features of the sim-
ulator is to provide un-patterned scenario generation
along with a learning objective and feedback as to a
learner’s defective operations from the lack of knowl-
edge on “project management principle”. Experimen-
tal results shows our proposed functionalities does
work and make sense from “problem solving-based
learning” viewpoints. Our future work is how to mod-
ify the scenario generation and feedback along with
the progress of a learner’s study.
This work was partially supported by KAKENHI;
Grant-in-Aid for challenging Exploratory Research
Davidovitch, L., Shtub, A., and Parush, A. (2007). Project
Management Simulation-Based Learning For Systems
Engineering Students. In Proceedings of Interna-
tional Conference on Systems Engineering and Mod-
eling 2007, ICSEM ’07, pp.17–23.
Drappa, A. and Ludewig, J. (2000). Simulation in software
engineering training. In Proceedings of the 22nd in-
ternational conference on Software engineering, ICSE
’00, pp.199–208.
Horine, G. (2005). Absolute Beginner’s Guide to Project
Management (Absolute Beginner’s Guide). Que
Corp., Indianapolis, IN, USA.
PMI (2009). The PMBOK Guide. Project Management In-
stitute, fourth edition.
The project status derived from your
operation at "module 4" of "project 2"
is shown below.
Which operation is the most proper one
to improve your operation?
Please choose it from the choices.
[Your operation]
overtime directive;
7th day, 10th day, 13th day
[Your result]
Developpment term;
1. 4th day; progress check
8th day; overtime directive
10th day; progress check
2.11th day; overtime directive &
progress check
12th day; progress check
3. 4th day; overtime directive
7th day; overtime directive &
progress check
4.12th day; overtime directive &
progress check
15th day; overtime directive