JISC Project Plan

siberiaskeinΔιαχείριση Δεδομένων

20 Νοε 2013 (πριν από 3 χρόνια και 8 μήνες)

182 εμφανίσεις

Project Acronym:
MCMS

Version:

0.1

Contact:

Samia Oussena samia.oussena@tvu.ac.uk

Date:
04/11/2008



Page
1

of
8

Document title: JISC Project Plan

Last updated: April 2007



JISC Project Plan



Overview of Project

1. Background


HE institutions collect vast amounts of data on students and courses which are not integrated and are
not easily queried or mined. Therefore relatively little data is turned into knowledge that
can be used
for institutional learning. The project aims to build a data management system that will use
knowledge discovery tools such as data mining in innovative ways based on the integration of a
number of institutional data sources such as e
-
learning

and e
-
administration. The objective is to
integrate key subsets of the data which would then be analyzed with knowledge discovery tools.

The majority of data which is accessible is quantitative. For example, most assessment data is
summative and numerical
. Combined with the fact that students often find assessment and
progression regulations complex. The project
propose to

use natural language generation techniques
as a method of creating personalized natural language reports


2. Aims and Objectives


The
aim of the project is to support the institution student retention strategy. The project will build a
system that will monitor student information sources and generates events, reports, advice to identify
potential divergence from and to recommend remedial

actions for prescribed educational processes.

The aims of the project will be met by the following objectives:



The project will conduct a detailed survey of the stakeholder’s main areas of concerns and
good intervention practices. Example questions inc
lude:

o

Who are the students who are most likely to drop out?

o

Who are the students most likely to benefit from e
-
Learning?

o

Which courses are most likely to benefit from e
-
Learning?

o

What hidden factors make for successful and independent learners?

o

Which cours
e has a high drop out?



The project will conduct a data analysis of the institution database systems relevant to the
problem areas.



The project will propose and implement a data integration model.



The project will build and evaluate data mining models.



The
project will build an application that will use the data mining model and implement
intervention requirements.



The project will evaluate the effectiveness of the proposed approach.

3. Overall Approach



The project will follow the CRISP_DM (Cross Industry
standard process for data mining) methodology.

Project Acronym:
MCMS

Version:
0.1

Contact:

Samia Oussena (samia.oussena@tvu.ac.uk)

Date:
03/11/2008


Page
2

of
8

Document title: JISC Project Plan

Last updated: April 2007





The Business Understanding Phase This will be an investigative process. The main
techniques that will be used in this phase are survey and interviews. The objective of the
investigation is to articulate probl
em areas that we will be addressing in this project in form of
questions relevant at the department, school, faculty and institution level.



The Data Understanding Phase This phase will require the reverse engineering of the
database systems that we will be

included in this study. We will be using both manual as well
as automatic techniques for the generation of the UML class diagrams. Following from the
data modelling of the existing database systems, we will propose a unified schema in a UML
class diagram
.



The Data Preparation Phase: Following from the previous phase, scripts and triggers will be
developed for the migration of data from the different sources to a unified repository. The
repository will be implemented in Oracle 11g. Here, the main data
preparation will include the
harmonisation of the data types across the different data sources, as well as the
depersonalisation of data when it is required.



The Modelling Phase This is the phase, where data mining algorithms sift through the data to
find
patterns and to build predictive models. We will be using both supervised and
unsupervised leaning algorithms. The algorithms provided with oracle miner
1

that will used in
this phase are: Apriori_association_rules, Decision_tree, generalized_linear_model,
K
-
means,
Naïve_bayes, NonNegative_matrix_factor, O_cluster, Support_vector_machines.



The Evaluation Phase: In this phase, we will be reviewing the different models constructed
previously and select the ones that will best suited to the project requirements
.



The Deployment Phase: This is the phase that makes use of the created models. The
required actions will be implemented in the personalised alerts. As shown in the architecture
diagram we will build a set of web services that will interface with the data

mining models
using a Java API and the text mining system. These services will be consumed by a
personalised alert system and a statistical report system. The later tool will be implemented
as a visualisation tool and provide periodic reports.



Project

Evaluation: We will develop an evaluation strategy that will allow us to gather
evidence to support areas such as efficiency of the approach, the extent of support of
institutional strategies, tangible benefits in terms of efficiencies and student perform
ance and
engagement. The evaluation plan will describe the audience for the evaluation, appropriate
evaluation questions, data collection and data analysis method



4. Project Outputs

The following outputs will be produced by the project:



Student in
educat
ion

context model
. In this model, we will concentrate on student information
relevant to students’ retention strategy implementation.



Course/programme model: We will be concentrating on course/programme information
relevant to the monitoring of course/pro
gramme success and its relationship to student
retention strategy implementation



Data models of the institution data sources: The databases that we will be modelling are
course DB, student record system, institution VLE and student access to library system
.



Models and scripts of data transformation and data integration
of the institution data sources.



Data mining models



Intervention process models and implementation. The implementation will include
personalised alerts system.



Project evaluation report



Final report


1.




1


http://www.oracle.com/technology/products/bi/odm/pdf/data
-
mining
-
11g
-
datasheet.pdf


Project Acronym:
MCMS

Version:
0.1

Contact:

Samia Oussena (samia.oussena@tvu.ac.uk)

Date:
03/11/2008


Page
3

of
8

Document title: JISC Project Plan

Last updated: April 2007



5. Project Outcomes

We anticipate the following project outcomes:



A common and integrated student retention intervention process
es



An integrated view of institution data related to students



Experience of implementing data mining system in
HE environment



6
. Stakeholder Analysis

<
List key stakeholder groups and individuals that will be interested in your project outcomes, will be
affected by them, or whose support/approval is essential, both within your institution and in the
community, and

assess their importance (low/medium/high).
>


Stakeholder

Interest / stake

Importance

JISC



Synthesis project

Synthesis of all project
outputs and coordinating
project exchange

High

Other JISC Projects / Programmes (cluster)

Output from the projects might
be relevant to this project and
our outputs might relevant to
them.

Medium/Low

Non
-
Institution external organisations



UK
national statistics body

Data
collected and presented
is relevant to the project

Medium/high

Institutions



Senior Management

Support is essential to the
project. Project outcomes
relevant to the future strategy

High

F
aculties
senior management

(Dean, Teaching
associate dean)

Support essential to data
gathering and analysis;

Study results will be highly
faculty strategy

High

Institution data source owners

Support essential to data
gathering and analysis;


High

Faculties staff

Support essential to data
gathering and analysis;

Project will have impact on
their involvement in
student
retention intervention

High

Students

Student perspective of
the
institution support is highly
relevant
;

project

results will have an
indirect effect on the student
experience

High

7. Risk Analysis


Risk

Probability

(1
-
5)

Severity

(1
-
5)

Score

(P x

S)

Action to Prevent/Manage Risk

Project schedules slippage

4

2

8

Regular project management
meetings and plan update; Use of
virtual meetings (Skype)

Not enough data

1

5

8

Include in the integration of any
Project Acronym:
MCMS

Version:
0.1

Contact:

Samia Oussena (samia.oussena@tvu.ac.uk)

Date:
03/11/2008


Page
4

of
8

Document title: JISC Project Plan

Last updated: April 2007



systems that are relevant and have
information on students and
courses.

Access to stakeholders

2

3

6


Use a number of communication
methods (email, phone, visit)

Recruitment of Research
Assistant on short term
contract

2

3

6


Consider using internal resources
/ expertise differently.

Loss of key personnel

2

3

6

Ensure that there is overlap
between roles

Project Team availability

3

2

6

Ensure that line managers of
project team are aware of the
project and the resource
requirements throughout the
project life cycle.


8. Standards



Name of standard or
specification

Version

Notes

UML

UML 2.0

Will be using class
diagram and activity
diagram model

BPMN

1.0

Will be used in the implementation if it is
relevant

Oracle data mining engine

Oracle 11g

Maturity of the tool. We will also
review open
source engine.

CRISP_DM (Cross Industry
standard process for data
mining) methodology



XML


All fusion data will expressed in XML. We
will try to generate the schema from the data
model.





9. Technical Development

We

will utilise UML and other modelling notations such as BPMN to capture process descriptions,
structural information relating to student, course/programme, data sources and intervention
processes
.

The project will follow an MDA approach when it is possible
.

Any other software development will
follow an iterative process. Relevant IDE, framework will be selected.

The project will define and agree a change control and configuration management process early in the
project lifecycle. The process will address t
he following points: Software and documentation will be
based
-
lined and be placed under version control using Source Safe

Changes to software, documentation and other artefacts will be controlled, authorised and traceability
between changes will support au
diting of the processes.

S
oftware produced will have an accompanying Test specification and Test Plan. Problems reported
during testing will be recorded using a standard bug reporting
Template and will be
auditable against
changes to the software as per
the change control and configuration management processes.

All documentation produced, designs, report will be produced in MS word and other MS Office
products together with PDF versions, Models of design will be expressed in UML 2.0 and PDF
documents.

Project Acronym:
MCMS

Version:
0.1

Contact:

Samia Oussena (samia.oussena@tvu.ac.uk)

Date:
03/11/2008


Page
5

of
8

Document title: JISC Project Plan

Last updated: April 2007



10.

Intellectual Property Rights


The Oracle licensing for HE is changing, we will be looking at the implication of using the Oracle data
mining engine
within the

project life
time and

after

the first pilot.


The code will be published via the project website

and made available under an appropriate open
source agreement and may be used within any educational establishment in line with JISC’s
requirements, as per the terms and conditions of JISC grants. The University and its partners will
retain shared IPR on
the software artefact's, and associated documentation. This will be confirmed via
a Consortium agreement for defining IPR arrangements that conforms to JISC requirements.


Sustainability of the code produced is through ensuring other institutions and JISC/
CETIS projects
have access to the code and documentation for the system, through LGPL or GPL licenses. All
reports, tools and code from the project will remain on the project server for a minimum period of 3
years and archived in an appropriate JISC reposi
tory.


Project Resources

11. Project Partners


University of Birmingham: This partner will provide the expertise in the area of natural language
process. It will be used in the analysis phase for the text mining and in the personalised alert system.

Main
Contact

:
Dr. Mark Lee


12. Project Management


The project team will meet regularly to monitor progress across work packages, monitor and manage
risks, agree changes and address major issues.

There will be a weekly skype meeting mostly face to
face meetin
g of the core development team. There will be a face to face team meeting for each
milestone.


Day to day project management will be coordinated by the Project Leader.

The Project Management Board will use a set of instruments for documentation and manag
ement that
will be set out in the Project Plan but will include a Risk Register, a Quality Assurance Plan and an
Issues Log

We will also be using an online project management tool that will allow access to all
member of the team (
.http://www.dotproject.
net/
)


List all members of the project team, their roles, and contact details. Indicate the proportion of time
the project manager will spend on project management.

The project Team is as follow:


Member Name

Role

Contact

RA1 and RA2 to be appointed




Dr
Samia Oussena


Project manager

Samia.oussena@tvu.ac.uk


Prof Tony Clark

Project Investigator

Tony.clark@tvu.ac.uk


Dr John Moor

Project Investigator

JohnPT.Moore@tvu.ac.uk


Prof Lynne Dunckley

Project Investigator

Lynne.dunckley@tvu.ac.uk


Dr Mark Lee

Consultant

m.g.lee@cs.bham.ac.uk


Maggie Stephens

Consultant

Maggie.stephens@tvu.ac.uk

John Waters

Consultant

John.waters@tvu.ac.uk

Project Acronym:
MCMS

Version:
0.1

Contact:

Samia Oussena (samia.oussena@tvu.ac.uk)

Date:
03/11/2008


Page
6

of
8

Document title: JISC Project Plan

Last updated: April 2007



Martina Rohr

Consultant

Martina.rohr@tvu.ac.uk




13. Programme Support

<
Indicate if there are specific areas where you would like

support from the programme or programme
manager.
>


14. Budget

<
Use the
budget template

and attach the project budget as Appendix A. Explain any changes from
the budget in the agreed project proposal.
>


Detailed Project Planning

15. Workpackages

<
Use the
workpackages template

to plan the detailed project work and attach as Appendix B. Clearly
indicate project deliverables and reports (in
bold
), when they are due, phasing of workpackages, and
explain any dependencies. You may also attach a Gantt chart, di
agram, or flowchart to illustrate
phasing
.
>


16. Evaluation Plan


Timing

Factor to Evaluate

Questions to Address

Method(s)

Measure of Success

TBC

Student and course
model

Is this a good
representation of
student & course
information in the
context of
student
retention

Review
workshop

Report to the
main
stakeholders

Endorsement of the
model by the
stakeholders

TBC

Data sources model

Is this a good
representation of the
data sources

Review
workshop +
implementation
of data
warehouse

+ report to
data so
urce
owners

Endorsement of the
model by the
stakeholders

The model was one of
main source for
implementing the data
warehouse

TBC

Data fusion model

Is the model a good
representation of data
fusion (integration

Review
workshop

implementation
of data
warehouse

+ report to
data source
owners

Endorsement of the
model by the
stakeholders

The model was one of
main source for
implementing the data
warehouse

TBC

Data mining models

Is the data mining
model efficient in
detecting problems

Section the
historic
al data
to be used as
test data + run
a pilot

The prediction are
accurate


Project Acronym:
MCMS

Version:
0.1

Contact:

Samia Oussena (samia.oussena@tvu.ac.uk)

Date:
03/11/2008


Page
7

of
8

Document title: JISC Project Plan

Last updated: April 2007



TBC

Effectiveness of the
presentation layer

Are the generated
reports a good

(accurate)

representation of the
intervention

Review
+Survey +
interview

The reports were
w
ell
received

TBC

Effectiveness of
MCMS

Is the intervention
process supported by
MCMS effective

Process
monitoring as
part of the
process

Improvement in the
student retention

17.
Quality Plan


Output


Timing

Quality
criteria

QA method(s)

Evidence of
compliance

Quality
responsibilities

Quality tools

(if
applicable)

Dec 08

Quality of
the
Models

Use of a tool +
Review

Model report
generated witjout
errors

Technical
Manager

IDE used for
representing
the models
(IBM RSA)

July 09

Quality of
the
Software

Standard
documentation +

Testing

(review of
test cases +
running the test
cases),

Passing the r
eview
of test cases +
software passing
the test cases

Technical
manager

Use junit or
equivalent

unit test
framework

Nov
08
-
Mar10

Quality of
project
documents

Document review

Passing the
document review

Project Manager
















18. Dissemination Plan


Timing

Dissemination Activity

Audience

Purpose

Key Message

First Month
and
continuing
after

Web site and blog

General and
technical
audience

Awareness,
inform , engage
and promote

MCMS project
development

Each
milestone

Analysis report
,Implementation report
and pilot evaluation
report

TVU stakeholders

Report on the
project finding
after the analysis

MCMS
development and
feedback

Through
out
the project

Publish in relevant
conferences, Journals

IT, HE
communities

Publish relevant
project findings

Engage and
promote

MCMS
development








19. Exit and Sustainability Plans


Project Acronym:
MCMS

Version:
0.1

Contact:

Samia Oussena (samia.oussena@tvu.ac.uk)

Date:
03/11/2008


Page
8

of
8

Document title: JISC Project Plan

Last updated: April 2007



Project Outputs

Action for Take
-
up & Embedding

Action for Exit

All reports

Will be posted on the project web site

Access: TVU will host the
server

Maintenance The server will
come under the maintenance
policy of the university

Software implementation

The code developed by the project
will be freely available to any
higher
Education institution.

TVU will host the code for
downloading

Maintenance The system will be
free to use by HE
establishments . All supporting
documentation , user manuals
will be freely available via the
web site. No on going
maintenance will be av
ailable
after the project closing date.




Project Outputs

Why Sustainable

Scenarios for Taking
Forward

Issues to Address

MCMS software
,
case studies, user
evaluations

C
a
n be used by other
JISC projects,
developers an
researcher


Using the software by
changing the data
integration layer

Review the data mining
model and check that it
is relevant in the new
context

Ap
pendixes

Appendix A. Project Budget

Appendix B. Workpackages