Indonesian J
ournal of Ele
c
trical Engin
eering and
Computer Sci
e
nce
Vol. 2, No. 2,
May 2016, pp
. 469 ~ 477
DOI: 10.115
9
1
/ijeecs.v2.i2.pp46
9-4
7
7
469
Re
cei
v
ed
Jan
uary 25, 201
6
;
Revi
sed
Ap
ril 20, 2016; Accepted Ma
y
1, 2016
Development of a Decision-Making System for Sultan
Moulay Slimane University in Beni Mellal,
Morocco
Abdellah
Amine*, Rachid
Ait Da
oud, Belaid Bouikh
alene
Sultan Mo
ul
a
y
Sliman
e Un
iver
sit
y
, F
a
cult
y
of Scienc
e an
d T
e
chn
o
lo
g
y
Po Bo
x 52
3, Beni Me
lla
l, Morocco
Phon
e: +
212 (0) 661
74
852
0 F
a
x: +
212 (0) 5
23 48
13
51
*Corres
p
o
ndi
n
g
author, e-ma
i
l
: a.amine
@
us
ms.ma
A
b
st
r
a
ct
The issue dealt with in this ar
ticle is to deve
lop a
decis
ion-m
a
k
i
ng info
r
m
ation system
r
e
lated t
o
the di
gita
l env
ir
on
me
nt of the
Univer
s
i
ty w
o
rk. W
e
propos
e t
o
mod
e
l th
e d
a
t
a w
i
thin the
u
n
iversity
in or
d
e
r
to
transfor
m
a system
of
inf
o
rm
ation into
a decision
-
m
ak
ing infor
m
ation sy
stem
,t
hat is
based on
thetrades
databases
oriented toward th
e actors. A decision-
making information
is a system
that allows the decis
ion
mak
e
rs of the
univ
e
rsity to h
a
ve rel
e
va
nt in
formati
on
a
nd
pow
erful a
naly
t
ical tools to h
e
lp the
m
tak
e
the
right dec
isio
n a
t
the right time.
Ke
y
w
ords
:
M
e
ta dat
a, strategic
infor
m
ati
o
n syste
m
s, us
er
classific
a
tio
n
mod
e
l, d
a
ta
w
a
rehous
e,
Meta
mo
de
lli
ng, Sql
Server
Copy
right
©
2016 In
stitu
t
e o
f
Ad
van
ced
En
g
i
n
eerin
g and
Scien
ce. All
rig
h
t
s reser
ve
d
.
1. Introduc
tion
De
cisi
on-ma
king an
d its executio
n
con
s
titu
te th
e funda
ment
al pu
rpo
s
e
s
of any
orga
nization
and
any ma
n
ageme
n
t. Ind
eed, a
n
y or
g
anization
dep
end
s
stru
cturally on th
e n
a
t
ure
of the de
ci
si
ons which
a
r
e taken by it
s d
e
ci
sion
-m
ake
r
s. The
d
e
ci
sion
-ma
k
e
r
s are
mo
re and
more
confron
t
ed with
ne
w
situation
s
an
d with
an
env
ironm
ent whi
c
h qui
ckly evolves.
In su
ch
a
context, de
ci
sion
-ma
k
in
g
has be
com
e
extremely di
ff
icult. So, the
intuitions an
d expe
rien
ce
of
the deci
s
io
n-make
rs are n
o
longe
r sufficient for
m
a
king the rig
h
t deci
s
io
ns.
In fact, the use
of
deci
s
io
n sup
port
system
s is vital to de
termine
th
e congruent info
rmation fo
r d
e
ci
sion
-ma
k
in
g.
They have p
r
oven to be a
n
essentia
l
e
l
e
ment that al
lows estim
a
ting the differe
nt situation
s
, the
v
a
riou
s ch
oic
e
s a
s
well a
s
t
heir impa
ct
s
[
1
]
.
2. The Busin
ess Intelligence Platform
The deci
s
io
n-makin
g
info
rmation syste
m
is
a set of
data organi
zed
in
a sp
eci
f
ic
way,
easily
acce
ssible a
nd
app
ropriate
to the
de
cisi
on-ma
k
i
ng;it is
in fac
t
an intelligent repres
entatio
n
of these data
throug
h sp
eci
a
lize
d
tools.
De
cisi
on Su
p
port Syste
m
s, at the level
of inte
ra
ct
ion
wit
h
t
h
e
de
ci
sion
ma
ke
r,
a
llow t
h
e
refund
and t
he an
alysis
of data from
different so
urces
ba
sed
on the te
ch
nologi
es of
mass
stora
ge, nam
ely the Data Wa
reho
usi
n
g
and OLAP.
Two mai
n
fun
c
tion
s are int
ende
d for the
deci
s
ion
-
ma
king tool
s (Fi
gure 1
)
:
1. Colle
cting
and
stori
n
g
,
ETL (Extra
cts, T
r
an
sforms, Lo
ad
s)
[2], Data Wareh
o
u
s
ing [
3
],
Datamart, dataweb.
2.
Extracting an
d pre
s
entin
g Data Minin
g
, OLAP.
2.1. Dev
e
lop
m
ent Tools
The BI pl
atform shoul
d p
r
ovide a
set o
f
prog
ramm
atic
devel
opme
n
t
tools and a
visua
l
developm
ent environ
ment,
couple
d
with
a software
d
e
velope
r’s
kit for creatin
g BI application
s
,
in ord
e
r to int
egrate th
em i
n
to a bu
sine
ss proc
ess, an
d/or emb
ed t
hem in an
oth
e
r ap
plication.
The BI
platform shoul
d al
so en
able
dev
elope
rs to
bui
ld BI ap
plications with
out
coding
by u
s
in
g
wizard
-li
k
e co
mpone
nts for
a grap
hical a
s
sembly process. [4]
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 25
02-4
752
IJEECS
Vol.
2, No. 2, May 2016 : 469 –
477
470
Figure 1. Architecture of the Busine
ss Intelligen
ce Pla
tform
2.2. Introduc
tion to Data
Mining
The term
Dat
a
Mining literally means
drilling dat
a. As in any drillin
g, its aim is to be able
to extract an
element: kn
owle
dge. Its con
c
e
p
ts a
r
e ba
sed o
n
the re
cog
n
ition within
ea
ch
comp
any of informatio
n hi
dden in the
d
a
ta rep
o
sito
ry. It allows, Th
anks to a nu
mber of
spe
c
i
f
ic
techni
que
s, it
allo
ws ma
ki
ng
kno
w
le
dg
e ap
pea
r. T
h
e Data Mi
nin
g
is the
set o
f
techni
que
s
and
method
s aim
ed at desig
n
ed for the exploratio
n
and
analysi
s
of large
com
put
er datab
ases in
orde
r to dete
c
t in these d
a
ta rule
s, asso
ciation
s
, u
n
kn
own tren
ds (not fixed a prio
ri), sp
ecific
stru
ctures, th
us
returning
in a con
c
ise
manne
r
the
essential i
n
formatio
n u
s
e
f
ul for de
ci
si
on
s
u
pport [5-7].
2.3. The Alg
o
rithms of
Data Mining
2.3.1. Classification
To ove
r
co
me
the limit of t
he bin
a
ry
cla
ssifi
cation i
n
respon
se
te
mplates,
Ch
a
r
les et al.
have ap
plied
ada
-bo
o
st [8
] to the Baye
sian
network
s
[9] [10]
s
u
ch as algorithms
for effec
t
ive
prog
ram
m
ing
to classify cu
stome
r
s
by th
eir aptitude to
resp
ond to bi
ds.
2.3.2. Rules
of As
socia
t
ions
The extra
c
tio
n
of rules of
asso
ciation
s
is
one of the tech
niqu
es of popul
ar data
mining,
whi
c
h
aims t
o
find th
e
rel
a
tionship
bet
wee
n
two
set
s
of
an
obje
c
t
.
The n
a
me
s
of the d
e
velo
pers
of algorithm
s
kno
w
n in thi
s
area a
r
e Ag
ra
wal et al
. [11] [12], Zak
i
Éc
lat [13], Han et al. [14].
2.3.3. Cluste
ring
The techniqu
e of cluste
ri
ng allo
ws toi
dent
ify the group
s of indi
viduals
with simila
r
c
h
arac
teris
t
ics
.
It c
a
n be us
ed to dis
t
inguis
h
s
e
gments
of
s
t
udents
by s
e
x /ins
titution/region... To
achi
eve thi
s
goal
seve
ral
algorith
m
s ha
ve bee
n u
s
e
d
su
ch
a
s
k-m
ean
s [15], fu
zzy c-me
an
s [
16],
the algorith
m
of Gustafson
-
Kessel [17] [
18] and the alg
o
rithm of Gat
h
Geva [19].
2.3.4. Prediction
Predi
cting th
e identity of one thin
g ba
sed
pur
ely o
n
the de
scrip
t
ion of anoth
e
r related
thing itba
sed
on the
relati
onship b
e
twe
en a thin
g th
at you can
know
and
a th
ing you n
eed
to
predi
ct [20]
3. Theore
t
ic
al Stud
y
of Our Applica
t
ion
In this
pa
rt, we
pre
s
e
n
t t
he
cla
s
s dia
g
r
am
and
the
seq
uen
ce
for the
con
s
truction of a
data wareho
use th
at ha
s
data from
the
univers
ity ap
plicatio
n kno
w
n a
s
APO
G
EE (Application
for the
o
r
ga
nization
an
d m
anag
ement
o
f
stude
nts an
d tea
c
he
rs). I
n
effect, th
e
obje
c
tive of t
h
is
work isto
ma
kea
n
a
ppli
c
ati
on u
nde
r the
Java
pro
g
ram
m
ing e
n
viro
n
m
ent in
orde
r to evalu
a
te t
h
e
algorith
m
s
of
data mini
ng o
n
data
set
s
th
at we
re
mi
ne
d at the level
of the data
wareh
o
u
s
e. Th
e
aim is also to pre
s
ent the st
atistics that descr
ib
es the
data wa
reh
o
u
s
e u
s
ing g
r
ap
hs an
d ch
arts.
3.1. The Clas
s Diagram
The cla
s
s dia
g
ram i
s
com
p
ose
d
of:
T
h
e Stud
e
n
t class
: It contain
s
the in
formation of a
ll
student
s en
rolled in the i
n
stitution.
Evaluation Warning : The document was created with Spire.PDF for Python.
IJEECS
ISSN:
2502-4
752
De
velo
pm
ent of a decisi
o
n
-
m
a
king
syst
em
for
Sultan Moulay Slim
ane … (Abd
ellah Am
ine)
471
T
h
e Insti
t
u
t
ion class
: It contai
ns the
name an
d the
acro
nym of the institution.
T
h
e Bacc
alaurea
t
e cl
ass
: It conta
i
ns the type and the seri
es of the st
udent'
s
ba
ch
elor
degree.
T
h
e Pa
th
class
: It contain
s
the
name, the
wordin
g of th
e ch
ain, the
diplom
aand
the
comp
one
nt or thestude
nt is registe
r
e
d
.
T
h
e Regio
n
class
: It contain
s
the country, the region,
an
d the city to which the
stud
ent
belon
gs.
Figure 2. Cla
ss di
agram of
our appli
c
ati
o
n
3.2. The Ac
tiv
i
t
y
Diagram
Activity diagrams a
r
e g
r
ap
hical rep
r
e
s
e
n
tations of workflo
w
s of st
epwi
s
e a
c
tivities an
d
action
s with
sup
port for
choice, iteratio
n and
co
n
c
u
r
ren
c
y. In the Unified Mo
d
e
ling La
ngu
a
ge,
activity diagrams a
r
e i
n
te
nded to m
o
d
e
l both
comp
utational a
n
d
orga
nization
al pro
c
e
s
se
s
(i.e.
workflows) [21]. Activity di
agrams
sh
ow the overall flow of cont
rol.
In this section, we will present
the Ac
tivity Diagram.
Figure 3. The
Activity Diagram
The
scenario of our
case
i
s
the following: The student f
ills out the enrollment form, the
tutor
che
c
ks
and
analy
z
e
s
that all
of t
he info
rm
atio
n
is
corre
c
t, and req
u
e
s
ts
the stude
nt to
confirm the
choice of the
chain. Th
e
stu
dent
conf
irms
the choi
ce of
the chain. T
he
tutor
records
and archive
s
the informatio
n of the
stude
nt and gives t
he stud
ent ca
rd.
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 25
02-4
752
IJEECS
Vol.
2, No. 2, May 2016 : 469 –
477
472
3.3. Applications of da
ta
mining
3.3.1. Classification
Cla
ssifi
cation
use
s
the de
cision tre
e
s, which offer a
n
outputthat is
clea
r and e
a
sy to interpret.
Cla
ssifi
cation
of students b
y
age.
Cla
ssifi
cation
of stu
dent
s
by sex: T
h
is
crit
e
r
ion
is u
s
ed to
determi
ne the
sex of
stud
ent to
make
clea
r th
e gend
er divi
sion a
nd its di
fferent rate
s.
Cla
ssifi
cation
of students b
y
secto
r
.
Cla
ssifi
cation
of students b
y
note.
Cla
ssifi
cation
of students b
y
region.
3.3.2. Algorithm used Clu
s
tering
K-me
ans
The al
gorith
m
of k-mea
n
s, p
r
e
s
ente
d
by McQu
een in
196
7
,
is an
algo
rithm fo
r
partitionin
g
of data falling within the stat
istics
and of the automati
c
prog
ram
m
ing
(more p
r
e
c
isely
of the
pro
g
ra
mming
whi
c
h
is not
su
pervised).
It is a
metho
d
who
s
e
pu
rpo
s
e
i
s
to
divide
th
e
observation
s
into K pa
rtitions
(cl
u
ste
r
s) i
n
whi
c
h
ea
ch
observatio
n
belon
gs to th
e pa
rtition wit
h
the clo
s
e
s
t averag
e [22].
3.3.3. Repre
sentation o
f
the Applicati
on
The appli
c
ati
on is divide
d into three pa
rt
s:
H
o
mep
a
ge
Data minin
g
part
S
t
at
ist
i
cs
p
a
rt
Figure 4. Interface of the A
pplication
3.3.3.1. Data
Mining Part
InData Min
g
p
a
rt, three al
g
o
rithm
s
we
re
used: Th
e cl
assificatio
n
, the clu
s
te
ring
and the
rule
s of
as
so
ciat
ion
s
.
The majo
rity of the work was
co
ncen
trated
on the
classification
of student
s
by their
results; there are two
statu
s
e
s
of
the results: valid and invalid.
Evaluation Warning : The document was created with Spire.PDF for Python.
IJEECS
ISSN:
2502-4
752
De
velo
pm
ent of a decisi
o
n
-
m
a
king
syst
em
for
Sultan Moulay Slim
ane … (Abd
ellah Am
ine)
473
Figure 5. Interface fo
r Data
Mining Appli
c
ation
The indi
cators use
d
in this classificatio
n
ar
e the regio
n
, the study disci
pline, the sex and
the type of baccalau
r
eate.
For exampl
e
,
the contrib
u
tors
who b
e
long tothe
regio
nofTa
dla
and wh
o h
a
ve a
bachelo
r
'
s
de
gree
with disti
n
ction h
a
ve the equivale
nt
of valid module 2.
3.3.3.2. Stati
s
tic Part
The stati
s
tical
part is divide
d into four top
i
cs:
Inscription
:
displ
a
ying stu
dents e
n
roll
e
d
by thei
r regi
ons a
nd Baccalaureate di
stinction
Validation
: displ
a
ying st
udent
s who
validated
b
y
t
heir i
n
stit
utions
an
d by
years of
grad
uation
Sex
: display
of student
s e
n
rolle
d by their sexe
s
Bours
e
: displ
a
y of student
s’ schol
arship
accordi
ng to their re
gion
s
Figure 5. Interface fo
r stati
s
tic ap
plicatio
n
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 25
02-4
752
IJEECS
Vol.
2, No. 2, May 2016 : 469 –
477
474
a.
The Inscripti
on Part
Figure 6. Reg
i
stration of Student
s by Re
gion
s
The majo
rity of student
s come from the
region
s of Ta
dla-A
z
ilal and
Chao
uia.
Figure 7. Reg
i
stration of Student
s by Bac
This pa
rt in
cl
ude
s fou
r
typ
e
s
of ho
nors:
TB
:
Very G
ood (Ist cla
s
s)B:
G
ood
(u
p
per
IInd
class) AB: good enough (Lower IInd cl
ass) P: passable (IIIrd class)
The majo
rity of student
s h
a
ve the degre
e
: very good und go
od en
o
ugh
b.
The v
a
lidation part
The Unive
r
sity Sultan Moulay Slimane is
comp
rised of
the following
institution
s
:
FLSH
: Faculty of Letters a
nd Hum
an Sciences
FST
: Faculty of Sciencean
d Tech
niqu
es
FP
: Multidisciplinary College
EST
: Higher
Colle
ge of Te
chn
o
logy
Evaluation Warning : The document was created with Spire.PDF for Python.
IJEECS
ISSN:
2502-4
752
De
velo
pm
ent of a decisi
o
n
-
m
a
king
syst
em
for
Sultan Moulay Slim
ane … (Abd
ellah Am
ine)
475
Figure 8. validation of Students by Fa
culty
Figure 9. validation of Students by year
The Fi
gures
8 and 9 illustrate
the number of
students wh
o have validated the ex
ams
by
Faculty and b
y
year.
c.
Sex of stude
nts par
t
Figure 10. Re
gistratio
n
of Students by se
x
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 25
02-4
752
IJEECS
Vol.
2, No. 2, May 2016 : 469 –
477
476
d.
The stu
d
en
ts fello
w
s
par
t
Figure 11. Th
e Students fel
l
ows by their regio
n
s
4. Conclusio
n
This a
r
ticl
e is intended to
put in pla
c
e
a De
ci
sion S
uppo
rt Syste
m
for the ma
nagem
ent
of stude
nts a
nd sho
w
s
ho
w it is p
o
ssibl
e
to expl
oit the existin
g
d
a
taba
se
s in o
r
de
r to de
sig
n
a
data wa
reh
o
u
s
e that integrates the de
si
gn and
the n
e
eds of en
d-u
s
ers.
Actors-orie
n
ted trad
es’ d
a
t
abases
all
o
w giving sug
g
e
s
tion
s in de
ci
sion
-ma
k
in
g. Finally,
we
have
de
si
gned
an
ap
pli
c
ation
un
der t
he
Java
prog
rammin
g
environm
ent in
o
r
der to eval
uate
the alg
o
rithm
s
of
data
mi
ning
on th
e
basi
s
of dat
aset
s that
were
extra
c
ted
from
the
da
ta
wareho
use. The arti
cle al
so pr
esents t
he statisti
cs
that describ
e
s
the data wareh
o
u
s
e u
s
i
n
g
grap
hs a
nd charts.
Referen
ces
[1]
Lévi
ne P, Pom
e
rol J. Editors.
Systèm
es
inter
a
ctifs d’
aide à
la
décis
ion et systèm
es
experts
. Hermès.
1
9
89
[2]
Golfarel
li M,
Rizzi S.
Ed
ito
r
s. Data W
a
re
hous
e D
e
sig
n
, Moder
n Pri
n
ciples
an
d Me
thodo
lo
gies
.
McGra
w
-
Hil
l Educati
on. 20
09
.
[3]
Inmon W.H. Editors. Build
ing t
he data
w
a
re
h
ouse.
Wile
y
. 1
996.
[4]
W
i
eder
B, Ossimitz M
. T
h
e
i
m
p
a
ct of
Busi
ness Int
e
lli
ge
n
c
e o
n
th
e
qua
l
i
ty of d
e
cis
i
on-
mak
i
n
g
–
a
me
di
ation M
o
d
e
l.
Conf
erenc
e
on ENT
E
Rpri
se Informatio
n
S
y
stems. CE
NT
ERIS. 2015; 64:
116
3–
117
1.
[5]
T
u
ffer
y
S. Ed
itors. Datami
ni
ng et
statisti
q
ue d
é
cisi
on
nel
le
: l'
intel
lig
en
ce des
do
nn
é
e
s.Editio
n
s
techni
p
. 201
2.
[6]
Ro
y
B, Bo
u
y
ss
ou D. Editors. Aide mu
lticritèr
e
à la déc
isi
on
:
métho
des et cas. Econo
mic
a
.199
3.
[7]
Sidd
artha .B.
Di
rect Marketing
Resp
onse Mo
d
e
ls usi
ng Gen
e
t
ic Algorith
m
s
.
America
n
Asso
ciatio
n for
Artificial Intel
lig
ence.1
9
9
8
: 14
4-14
8.
[8]
F
r
eund
Y, Sch
apir
e
RE.
Ex
pe
riments w
i
th
a
new
bo
ostin
g
a
l
gorit
hm
.
In Pro
c
eed
ings
of th
e
T
h
irteen
t
h
Internatio
na
l C
onfere
n
ce o
n
Machi
ne Le
arn
i
ng.1
9
9
6
: 148
– 156.
[9]
Elkan C.
Boosti
ng
a
nd naiv
e
Bayesi
an
l
ear
n
i
ng
. D
epartme
n
t of Comp
ute
r
Scienc
e an
d
Engi
neer
in
g
Univers
i
t
y
of Califor
nia. 19
97.
[10]
Lan
gle
y
P, Iba
W
,
T
hompson
K.
An Analysis of Bayes
i
an
Classifiers
. In
Procee
di
ngs
of the T
enth
Natio
nal C
onfe
r
ence o
n
Artificial Intel
lig
enc
e. Moffet F
i
eld. 1992: 22
3 – 22
8
.
[11]
Agra
w
a
l R, Imielinsk
i T
,
S
w
a
m
i A.
Mining
a
ssociati
on rul
e
s betw
een sets
of items i
n
lar
ge dat
abas
es
.
In Proceed
in
gs
of the ACM SIGMOD
Conference. W
a
shi
n
g
t
on.199
3: 207-
216.
[12]
Agra
w
a
l R, Srikant R .
Fast Al
gorithm
s for Mining Association Rules
. In Pro
c
eed
ings of the
20th VLD
B
confere
n
ce.Sa
n
tiag
o.19
94: 4
87-4
99.
[13]
Z
a
ki MJ. Scala
b
le
alg
o
rithms
for associ
atio
n
mini
ng.
IEEE
Transactions
on Knowledge and Data
Engi
neer
in
g
. 2000; 12(
3): 372
–39
0.
Evaluation Warning : The document was created with Spire.PDF for Python.
IJEECS
ISSN:
2502-4
752
De
velo
pm
ent of a decisi
o
n
-
m
a
king
syst
em
for
Sultan Moulay Slim
ane … (Abd
ellah Am
ine)
477
[14]
Han J, Pei J, Y
i
n Y.
M
i
ni
ng F
r
equ
ent P
a
ttern
s w
i
thout C
a
n
d
i
date
G
ener
ation
.SIGMOD Proceedings
of T
he ACM SIGMOD internatio
nal
confer
ence
on
M
a
n
a
geme
n
t of dat
a. Ne
w
Y
o
rk. 2
000; 2
9
(2): 1
-
12.
[15]
Kaufman
L, Ro
ussee
u
w
P J,
Dod
ge
Y. Stati
s
tical D
a
ta An
a
l
y
s
is Bas
ed
on
the L
1–N
orm
and
Rel
a
te
d
Methods.Els
evi
e
r Scienc
e.19
8
7
: 405–
41
6.
[16]
Dun
n
JC.
A
F
u
zz
y R
e
lativ
e
of the ISODA
T
A Process a
nd Its Us
e i
n
Detecti
ng
Co
mpact W
e
l
l
-
Separ
ated C
l
u
s
ters.
Journal of
Cybern
e
tics.
197
3; 3(3): 32-
57.
[17]
Gustafson D E., Kessel W C.
F
u
zz
y
C
l
usteri
ng w
i
th a F
u
zzy Covari
anc
e Matrix
, Decisi
o
n
an
d Co
ntro
l
Includ
ing t
he 1
7
th S
y
mpos
iu
m on Ad
aptiv
e
Proce
sses, IE
EE Confer
enc
e. S
an D
i
e
go.
197
8; 17:7
61-
766.
[18]
Babuk
a R, V
a
n der
Ve
en P
J
, Ka
y
m
ak U.
Impr
oved
cov
a
ria
n
ce
esti
ma
tion for Gust
afson-Kess
e
l
Clusteri
n
g
. IEEE Confere
n
ce.
Hon
o
lu
lu. 20
02
; 2: 1081-1
085.
[19]
Sar
w
ar B†,Kar
yp
is G‡, Konstan J†, Rie
dl J
†
.
Reco
mmend
er Systems for
Large-sc
ale E
-
Co
mmerce
:
Scala
b
le Nei
g
h
borh
ood
F
o
rmation Usin
g
C
l
u
sterin
g
.T
he Internatio
nal
Co
nferenc
e o
n
C
o
mputer
an
d
Information T
e
chno
log
y
. D
h
a
k
a. 2002.
[20]
Z
hang H, Z
h
a
o
G, Chen L, Li
u B. Short-T
e
rm Predi
ctio
n of
W
i
nd Po
w
e
r B
a
sed
on a
n
Improve
d
PSO
Neur
al N
e
t
w
or
k.
T
E
LKOMNIKA Indon
esi
an
Journ
a
l
of E
l
e
c
trical En
gin
e
e
rin
g
.
20
14
; 12(7): 4
9
7
3
-
498
0.
[21]
Rumb
aug
h J,
Jacobs
on I, B
ooch G.Ed
i
tor
s
.
T
he Un
ified
Mode
lin
g L
a
n
gua
ge
Refere
nce Ma
nu
al
.
Addis
on-W
e
sl
e
y
. 19
99.
[22]
Sai Ch
an
da
na
B, Srinivas K,
Kiran Kum
a
r
R.
Clusteri
ng
Algorit
hm Com
b
in
ed
w
i
t
h
Hi
ll
Climb
i
ng f
o
r
Classific
a
tio
n
of Remote S
ensi
ng Imag
e
.
Internation
a
l
Journa
l of
Electrical an
d
Co
mputer
Engi
neer
in
g (IJECE)
. 2014; 4(
6): 923-9
30.
Evaluation Warning : The document was created with Spire.PDF for Python.