I
AE
S In
t
er
na
t
io
na
l J
o
urna
l o
f
Art
if
icia
l In
t
ellig
ence
(
I
J
-
AI
)
Vo
l.
15
,
No
.
4
,
A
u
g
u
s
t
20
26
,
p
p
.
3
4
7
6
~
3
4
9
1
I
SS
N:
2
2
5
2
-
8
9
3
8
,
DOI
: 1
0
.
1
1
5
9
1
/ijai.v
15
.i
4
.
p
p
3
4
7
6
-
3
4
9
1
3476
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ij
a
i
.
ia
esco
r
e.
co
m
Co
mpa
riso
n alg
o
rithms wi
th
o
p
ti
miza
tion for cl
ust
ering
multi
-
c
riter
ia
I
da
M
uly
a
di
1
,
T
it
ik
K
ha
wa
Abdu
l R
a
hm
a
n
2
,
M
uh
a
m
m
a
d F
a
is
a
l
1
1
D
e
p
a
r
t
me
n
t
o
f
I
n
f
o
r
mat
i
c
s
,
U
n
i
v
e
r
si
t
a
s
M
u
h
a
m
ma
d
i
y
a
h
M
a
k
a
ss
a
r
,
M
a
k
a
ss
a
r
,
I
n
d
o
n
e
si
a
2
D
e
p
a
r
t
me
n
t
o
f
S
c
h
o
o
l
o
f
S
c
i
e
n
c
e
a
n
d
Te
c
h
n
o
l
o
g
y
,
A
si
a
E
U
n
i
v
e
r
s
i
t
y
,
K
u
a
l
a
L
u
mp
u
r
,
M
a
l
a
y
s
i
a
Art
icle
I
nfo
AB
S
T
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
Sep
3
,
2
0
2
5
R
ev
is
ed
J
u
n
5
,
2
0
2
6
Acc
ep
ted
J
u
l
9
,
2
0
2
6
In
h
ig
h
e
r
e
d
u
c
a
ti
o
n
,
fo
re
c
a
stin
g
st
u
d
e
n
t
g
ra
d
u
a
ti
o
n
is
c
ru
c
ial
fo
r
e
a
rly
in
terv
e
n
t
io
n
d
e
v
e
lo
p
m
e
n
t,
p
o
l
icy
fo
rm
u
latio
n
,
a
n
d
c
las
s
p
lan
n
in
g
.
T
h
is
c
a
teg
o
rize
s
th
e
fa
c
to
rs
th
a
t
a
ffe
c
t
stu
d
e
n
t
g
ra
d
u
a
ti
o
n
a
c
c
o
rd
i
n
g
to
b
o
th
a
c
a
d
e
m
ic
a
n
d
n
o
n
-
a
c
a
d
e
m
ic
traits.
To
c
o
m
p
a
re
v
a
rio
u
s
c
lu
ste
rin
g
tec
h
n
iq
u
e
s,
su
c
h
a
s
K
-
m
e
a
n
s
,
fu
z
z
y
C
-
m
e
a
n
s
(F
CM
),
K
-
m
e
d
o
id
s
,
d
e
n
sity
-
b
a
se
d
sp
a
ti
a
l
c
lu
ste
rin
g
o
f
a
p
p
l
ica
ti
o
n
s
wit
h
n
o
ise
(
DBSCAN
)
,
sp
e
c
tral
c
lu
ste
rin
g
,
G
a
u
ss
ian
m
ix
t
u
re
m
o
d
e
l
(
G
M
M
)
,
a
n
d
d
e
e
p
lea
rn
i
n
g
(D
L)
,
a
s
we
ll
a
s
p
a
rti
c
le
sw
a
rm
o
p
ti
m
iza
ti
o
n
(
PSO
)
a
n
d
g
e
n
e
t
ic
a
lg
o
rit
h
m
(
GA
)
o
p
ti
m
iza
ti
o
n
f
o
r
K
-
m
e
a
n
s
a
n
d
F
CM
,
th
is
a
rti
c
le
a
p
p
l
ies
th
e
h
y
b
r
i
d
e
lb
o
w
+
silh
o
u
e
tt
e
o
p
t
imiz
a
ti
o
n
p
rio
r
t
o
th
e
c
lu
ste
rin
g
p
ro
c
e
ss
.
Two
c
lu
s
ters
we
re
fo
u
n
d
i
n
th
e
p
re
-
c
lu
ste
r
o
p
ti
m
i
z
a
ti
o
n
re
se
a
rc
h
re
su
lt
s.
K
-
m
e
a
n
s
,
F
CM
,
G
M
M
,
a
n
d
DL
c
l
u
ste
rin
g
a
ll
d
e
m
o
n
stra
ted
m
o
re
d
isti
n
c
t
c
e
n
tr
o
id
se
p
a
ra
ti
o
n
;
K
-
m
e
a
n
s
a
n
d
G
M
M
we
r
e
th
e
m
o
st
v
isu
a
ll
y
sta
b
le
a
n
d
c
o
m
p
r
e
h
e
n
sib
le
.
In
ter
-
c
lu
ste
r
se
p
a
ra
ti
o
n
is
t
h
e
m
a
in
g
o
a
l
o
f
p
o
st
-
c
lu
ste
r
o
p
ti
m
iza
ti
o
n
,
a
n
d
t
h
e
K
-
m
e
a
n
s
+
P
S
O
m
e
th
o
d
is
th
e
b
e
st
o
p
ti
o
n
.
Th
e
c
ro
ss
-
d
a
tas
e
t
v
a
li
d
a
ti
o
n
fin
d
i
n
g
s
in
d
ica
te
th
a
t
t
h
e
c
lu
ste
ri
n
g
m
o
d
e
l
e
x
h
ib
it
s
m
o
d
e
ra
te
c
o
n
s
isten
c
y
o
n
th
e
n
e
w
d
a
tas
e
t,
with
a
n
a
d
ju
ste
d
Ra
n
d
in
d
e
x
(
ARI
)
o
f
0
.
4
7
8
a
n
d
a
n
o
rm
a
li
z
e
d
m
u
t
u
a
l
in
f
o
rm
a
ti
o
n
(
NMI
)
o
f
0
.
4
3
3
.
K
ey
w
o
r
d
s
:
E
lb
o
w
Fu
zz
y
C
-
m
ea
n
s
K
-
m
ea
n
s
Op
tim
izatio
n
Sil
h
o
u
ette
T
h
is i
s
a
n
o
p
e
n
a
c
c
e
ss
a
rticle
u
n
d
e
r th
e
CC B
Y
-
SA
li
c
e
n
se
.
C
o
r
r
e
s
p
o
nd
ing
A
uth
o
r
:
I
d
a
Mu
ly
ad
i
Dep
ar
tm
en
t o
f
I
n
f
o
r
m
atics,
U
n
iv
er
s
itas
Mu
h
am
m
ad
iy
a
h
Ma
k
ass
ar
Su
ltan
Alau
d
d
in
No
.
2
5
9
,
Gn
.
Sar
i,
R
ap
p
o
cin
i,
Ma
k
ass
ar
C
ity
,
So
u
th
Su
lawe
s
i
,
I
n
d
o
n
esia
E
m
ail: id
am
u
ly
ad
i@
u
n
is
m
u
h
.
ac
.
id
1.
I
NT
RO
D
UCT
I
O
N
Un
d
er
s
tan
d
in
g
th
e
f
ac
to
r
s
a
s
s
o
ciate
d
with
s
tu
d
en
t
g
r
ad
u
atio
n
is
an
im
p
o
r
tan
t
is
s
u
e
in
h
ig
h
er
ed
u
ca
tio
n
b
ec
au
s
e
it
s
u
p
p
o
r
t
s
ac
ad
em
ic
p
la
n
n
in
g
,
p
o
licy
d
ev
elo
p
m
en
t,
an
d
ea
r
l
y
in
te
r
v
en
tio
n
s
tr
ateg
ies.
E
d
u
ca
tio
n
al
d
atasets
ar
e
ty
p
i
ca
lly
m
u
ltid
im
en
s
io
n
al
a
n
d
c
o
n
tain
co
m
p
lex
r
elatio
n
s
h
ip
s
am
o
n
g
v
ar
ia
b
les.
C
lu
s
ter
in
g
tech
n
iq
u
es
ar
e
m
a
ch
in
e
lear
n
in
g
m
eth
o
d
s
u
s
ed
t
o
ca
teg
o
r
ize
r
elate
d
d
ata
p
o
in
ts
with
o
u
t
th
e
n
ee
d
f
o
r
p
r
ed
eter
m
in
ed
class
lab
els.
K
-
m
ea
n
s
an
d
f
u
zz
y
C
-
m
ea
n
s
(
FC
M)
ar
e
f
r
eq
u
e
n
tly
u
s
ed
in
e
d
u
ca
tio
n
al
an
aly
tics
f
o
r
th
eir
v
er
s
atility
an
d
ef
f
ec
tiv
e
n
ess
ac
r
o
s
s
d
iv
er
s
e,
h
eter
o
g
e
n
eo
u
s
d
atasets
[
1
]
–
[
5
]
.
K
-
m
ea
n
s
is
o
f
ten
u
s
ed
f
o
r
its
co
m
p
u
tatio
n
al
ef
f
icien
cy
an
d
s
tr
aig
h
tf
o
r
war
d
im
p
le
m
en
tatio
n
.
No
n
eth
eless
,
its
ef
f
icac
y
d
ep
e
n
d
s
o
n
th
e
c
h
o
ice
o
f
in
itial
ce
n
tr
o
id
s
.
FC
M
is
m
o
r
e
ap
p
r
o
p
r
iate
f
o
r
o
v
e
r
lap
p
in
g
d
atasets
,
p
er
m
itti
n
g
d
ata
to
b
e
in
co
r
p
o
r
ated
in
to
n
u
m
er
o
u
s
clu
s
ter
s
.
Dete
r
m
in
in
g
th
e
n
u
m
b
e
r
o
f
clu
s
ter
s
b
ef
o
r
e
th
e
clu
s
ter
in
g
p
r
o
ce
d
u
r
e
y
ield
s
h
ig
h
ly
ac
cu
r
ate
r
esu
lts
[
6
]
.
T
h
e
e
lb
o
w
tech
n
iq
u
e
s
er
v
es
as
a
p
r
elim
in
ar
y
r
ef
e
r
en
ce
;
h
o
we
v
er
,
th
e
elb
o
w
p
o
in
t
m
ay
n
o
t
alwa
y
s
b
e
d
is
tin
ctly
ap
p
a
r
en
t.
T
h
e
s
ilh
o
u
ette
ap
p
r
o
ac
h
en
h
an
ce
s
th
e
e
lb
o
w
m
eth
o
d
'
s
ef
f
icac
y
b
y
an
aly
zin
g
clu
s
ter
q
u
ality
th
r
o
u
g
h
in
tr
a
-
cl
u
s
ter
co
h
esiv
en
ess
an
d
in
ter
-
cl
u
s
ter
s
ep
ar
atio
n
[
7
]
.
T
h
e
task
o
f
g
r
o
u
p
i
n
g
g
r
ad
u
atio
n
-
r
elate
d
cr
iter
ia
s
h
a
r
es
s
im
ilar
ities
wi
th
s
eg
m
en
tatio
n
p
r
o
b
lem
s
in
h
ea
lth
ca
r
e,
b
u
s
in
ess
,
an
d
s
o
cial
an
aly
tics
.
Ac
r
o
s
s
th
ese
d
o
m
ain
s
,
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J Ar
tif
I
n
tell
I
SS
N:
2252
-
8
9
3
8
C
o
mp
a
r
is
o
n
a
lg
o
r
ith
ms w
ith
o
p
timiz
a
tio
n
fo
r
clu
s
teri
n
g
mu
lti
-
crit
eria
(
I
d
a
Mu
lya
d
i
)
3477
r
esear
ch
er
s
m
u
s
t
ad
d
r
ess
h
et
er
o
g
en
e
o
u
s
d
ata
ch
ar
ac
ter
is
tics
,
b
o
u
n
d
ar
y
ca
s
es,
an
d
th
e
n
ee
d
to
g
en
er
ate
clu
s
ter
s
th
at
ar
e
b
o
th
m
ea
n
in
g
f
u
l a
n
d
ac
tio
n
ab
le
f
o
r
d
ec
is
io
n
m
ak
er
s
.
Pre
v
io
u
s
s
tu
d
ies
r
ev
ea
l
r
esea
r
ch
g
ap
s
,
p
ar
ticu
lar
ly
t
h
e
e
m
p
lo
y
m
e
n
t
o
f
r
estricte
d
alg
o
r
ith
m
s
,
as
ev
id
en
ce
d
b
y
[
8
]
–
[
1
0
]
,
wh
ic
h
co
m
p
a
r
es
K
-
m
ea
n
s
,
K
-
m
e
d
o
i
d
s
,
an
d
FC
M
alg
o
r
ith
m
s
f
o
r
clu
s
ter
in
g
jo
b
s
with
o
u
t
in
clu
d
in
g
f
u
r
th
er
o
p
tim
izatio
n
s
to
en
h
an
ce
th
e
ac
cu
r
ac
y
o
f
th
ese
alg
o
r
it
h
m
s
in
clu
s
ter
in
g
m
u
lti
-
cr
iter
ia
d
ata.
T
h
is
r
esear
ch
en
h
a
n
ce
s
th
e
ac
cu
r
ac
y
an
d
ef
f
icien
cy
o
f
clu
s
ter
in
g
,
p
ar
tic
u
lar
ly
in
m
a
n
ag
in
g
m
u
lti
-
cr
iter
ia
d
ata,
b
y
in
co
r
p
o
r
atin
g
h
ith
er
to
u
n
u
tili
ze
d
m
eta
h
eu
r
is
tic
o
p
tim
izatio
n
s
tr
ateg
ies
[
1
1
]
.
Dete
r
m
in
e
th
e
id
ea
l
n
u
m
b
e
r
o
f
clu
s
ter
s
b
y
u
s
in
g
th
e
e
lb
o
w
tec
h
n
iq
u
e
o
r
g
ap
s
tatis
tic,
as
o
u
tlin
ed
by
[
1
2
]
.
T
h
is
r
esear
ch
ad
d
r
ess
es
th
e
am
b
ig
u
ity
en
c
o
u
n
ter
ed
in
d
eter
m
i
n
in
g
th
e
el
b
o
w
p
o
in
t
b
y
ap
p
ly
in
g
s
ilh
o
u
ette
o
p
tim
izatio
n
ap
p
r
o
ac
h
es
to
en
h
a
n
ce
th
e
s
elec
tio
n
o
f
th
e
n
u
m
b
e
r
o
f
cl
u
s
ter
s
.
Prio
r
s
tu
d
ies
also
ad
d
r
ess
m
u
lti
-
cr
iter
ia
clu
s
ter
in
g
as
s
h
o
wn
in
[
1
3
]
,
c
o
n
ce
n
tr
ated
s
o
lely
o
n
clu
s
ter
i
n
g
d
ata
u
s
in
g
a
lim
ited
n
u
m
b
er
o
f
cr
iter
ia.
T
h
is
r
esear
ch
ex
am
in
es
m
u
lti
-
cr
iter
ia
d
ata
clu
s
ter
in
g
,
in
co
r
p
o
r
atin
g
m
u
ltip
le
f
ac
to
r
s
to
ass
es
s
s
tu
d
en
t
g
r
ad
u
atio
n
an
d
ac
ad
em
ic
p
er
f
o
r
m
an
ce
,
s
o
o
f
f
er
in
g
a
m
o
r
e
h
o
lis
tic
an
d
p
er
tin
en
t a
p
p
r
o
ac
h
t
o
ed
u
ca
tio
n
al
r
ea
liti
es.
T
h
is
r
esear
ch
is
n
o
v
el
in
in
co
r
p
o
r
atin
g
elb
o
w
–
s
ilh
o
u
ette
o
p
tim
izatio
n
to
ascer
tain
th
e
clu
s
ter
lay
o
u
t
p
r
io
r
to
clu
s
ter
in
g
.
A
co
m
p
ar
is
o
n
was
co
n
d
u
cted
am
o
n
g
s
ev
er
al
clu
s
ter
in
g
tech
n
i
q
u
es,
i
n
clu
d
in
g
K
-
m
ea
n
s
,
FC
M,
K
-
m
ed
o
id
s
,
d
en
s
ity
-
b
as
ed
s
p
atial
clu
s
ter
in
g
o
f
ap
p
lic
atio
n
s
with
n
o
is
e
(
DB
SC
AN
)
,
s
p
ec
tr
al
clu
s
ter
in
g
,
G
au
s
s
ian
m
ix
tu
r
e
m
o
d
els
(
GM
M)
,
an
d
d
ee
p
lear
n
in
g
(
DL
)
[
1
4
]
–
[
1
6
]
.
T
h
e
clu
s
ter
in
g
o
u
tc
o
m
es f
r
o
m
K
-
m
ea
n
s
an
d
FC
M
ar
e
im
p
r
o
v
ed
u
s
in
g
p
ar
ticle
s
war
m
o
p
tim
izatio
n
(
PSO
)
an
d
g
en
etic
alg
o
r
ith
m
(
GA
)
to
en
h
an
c
e
clu
s
ter
q
u
ality
[
1
7
]
.
T
h
e
p
r
im
a
r
y
d
o
n
atio
n
s
o
f
th
is
r
esear
ch
a
r
e:
−
A
co
m
p
ar
is
o
n
a
n
aly
s
is
o
f
K
-
m
ea
n
s
an
d
FC
M
f
o
cu
s
in
g
o
n
f
ac
to
r
s
in
f
lu
en
cin
g
s
tu
d
en
t
g
r
ad
u
atio
n
.
−
E
v
alu
atio
n
o
f
ea
ch
alg
o
r
ith
m
'
s
p
er
f
o
r
m
an
ce
in
ter
m
s
o
f
clu
s
ter
s
ep
ar
atio
n
,
o
v
er
l
ap
h
a
n
d
lin
g
,
co
m
p
u
tatio
n
al
e
f
f
icien
cy
,
an
d
in
ter
p
r
etab
ilit
y
with
in
an
ed
u
c
atio
n
al
co
n
tex
t.
−
E
lb
o
w
m
eth
o
d
f
o
r
d
eter
m
i
n
in
g
clu
s
ter
q
u
a
n
tity
.
−
I
n
teg
r
atio
n
o
f
m
eta
h
eu
r
is
tic
o
p
tim
izatio
n
tech
n
iq
u
es to
en
h
a
n
ce
th
e
ac
cu
r
ac
y
o
f
clu
s
ter
in
g
.
2.
M
E
T
H
O
D
T
h
e
clu
s
ter
in
g
p
r
o
ce
s
s
,
wh
ich
b
eg
in
s
with
ac
ad
em
ic
an
d
n
o
n
-
ac
ad
em
ic
d
atasets
,
is
d
ep
icted
in
th
e
r
esear
ch
d
esig
n
in
Fig
u
r
e
1
.
E
l
b
o
w
o
p
tim
izatio
n
u
tili
zin
g
s
ilh
o
u
ette
an
aly
s
is
to
d
eter
m
in
e
t
h
e
id
ea
l
n
u
m
b
er
o
f
clu
s
ter
s
.
T
h
e
o
p
tim
izatio
n
o
u
tco
m
es
ar
e
ap
p
lied
in
K
-
m
ea
n
s
,
FC
M,
K
-
m
ed
o
id
s
,
DB
S
C
AN,
s
p
ec
tr
al
clu
s
ter
in
g
,
GM
M,
an
d
DL
[
1
8
]
.
T
o
m
a
k
e
s
u
r
e
th
e
clu
s
ter
in
g
r
esu
lts
ar
e
co
n
s
is
ten
t
an
d
d
ep
e
n
d
ab
le,
alg
o
r
ith
m
s
th
en
o
p
tim
ized
u
s
in
g
PS
O
an
d
GA,
ass
ess
ed
u
s
in
g
s
ilh
o
u
ette
s
co
r
e,
p
u
r
ity
,
n
o
r
m
alize
d
m
u
tu
al
in
f
o
r
m
atio
n
(
NM
I
)
,
an
d
ad
ju
s
ted
R
an
d
in
d
ex
(
AR
I
)
,
an
d
v
alid
ated
u
s
in
g
b
o
o
ts
tr
ap
p
in
g
an
d
cr
o
s
s
-
v
alid
a
tio
n
.
Fig
u
r
e
1
.
R
esear
ch
m
eth
o
d
o
l
o
g
y
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8
9
3
8
I
n
t J Ar
tif
I
n
tell
,
Vo
l.
15
,
No
.
4
,
Au
g
u
s
t
20
26
:
3
4
7
6
-
3
4
9
1
3478
2
.
1
.
Da
t
a
c
o
llect
io
n
Sam
p
les
f
r
o
m
ac
ad
em
ic
d
atasets
at
s
ev
er
al
u
n
iv
er
s
ities
ar
e
u
s
ed
in
th
is
s
tu
d
y
an
d
ar
e
s
u
p
p
o
r
ted
b
y
r
ef
er
en
ce
s
t
o
r
ele
v
an
t
r
esear
c
h
.
Dir
ec
t
o
b
s
er
v
atio
n
o
n
ca
m
p
u
s
an
d
a
r
ev
iew
o
f
th
e
liter
atu
r
e
wer
e
u
s
ed
to
g
ath
er
d
ata.
T
h
e
cr
iter
ia
wer
e
estab
lis
h
ed
b
y
r
ef
er
en
ci
n
g
p
r
e
v
io
u
s
r
esear
ch
in
th
e
d
o
m
ain
o
f
ed
u
ca
tio
n
al
d
ata
m
in
in
g
,
wh
ic
h
d
em
o
n
s
tr
ated
t
h
at
th
ese
f
ac
to
r
s
h
av
e
a
s
ig
n
if
i
ca
n
t im
p
ac
t o
n
s
tu
d
en
t g
r
ad
u
at
io
n
r
ates
[
1
9
]
.
T
h
e
p
r
o
p
o
s
ed
d
ataset
in
clu
d
es
3
3
c
r
iter
ia
th
at
r
ep
r
esen
t
co
n
tex
tu
al
ac
a
d
em
ic
a
n
d
n
o
n
-
ac
ad
em
ic
p
er
f
o
r
m
an
ce
v
ar
iab
les
ass
o
cia
ted
with
s
tu
d
e
n
t
ac
h
ie
v
em
en
t
[
2
0
]
,
[
2
1
]
.
T
h
e
c
u
m
u
lativ
e
g
r
ad
e
p
o
in
t
a
v
er
ag
e
(
GPA)
[
2
2
]
,
co
u
r
s
e
co
m
p
letio
n
r
ates,
th
e
n
u
m
b
e
r
o
f
r
e
p
ea
te
d
co
u
r
s
es,
s
em
ester
cr
ed
it lo
ad
,
an
d
g
en
e
r
al
tr
en
d
s
in
ac
ad
em
ic
ac
h
iev
em
en
t
[
2
3
]
–
[
2
6
]
ar
e
e
x
am
p
les
o
f
ac
ad
em
ic
f
ac
to
r
s
.
C
o
n
v
er
s
el
y
,
in
v
o
lv
em
en
t
in
ex
tr
ac
u
r
r
icu
la
r
ac
tiv
ities
,
s
ch
o
o
l
atten
d
an
ce
,
s
o
cio
ec
o
n
o
m
i
c
cir
cu
m
s
tan
ce
s
,
an
d
d
em
o
g
r
ap
h
ic
tr
aits
s
u
ch
as
ag
e
an
d
g
e
n
d
er
[
2
7
]
ar
e
e
x
am
p
les
o
f
n
o
n
-
ac
a
d
em
ic
in
f
lu
e
n
c
es.
T
h
ese
ch
ar
ac
ter
is
tics
wer
e
s
elec
ted
b
ec
au
s
e
a
n
u
m
b
er
o
f
i
n
ter
r
elate
d
f
ac
t
o
r
s
af
f
ec
t stu
d
en
t p
e
r
f
o
r
m
an
ce
an
d
g
r
ad
u
atio
n
p
r
o
s
p
ec
ts
.
Af
ter
th
at,
all
d
ata
v
al
u
es
ar
e
s
tan
d
ar
d
ized
to
th
e
s
am
e
r
an
g
e,
0
t
o
1
,
u
s
in
g
m
in
-
m
ax
n
o
r
m
aliza
tio
n
.
Fo
r
d
is
tan
ce
-
b
ased
clu
s
ter
in
g
tech
n
iq
u
es
lik
e
K
-
m
ea
n
s
an
d
FC
M,
th
is
s
tep
is
cr
u
cial.
1
6
s
p
ec
ialis
ts
ex
am
in
ed
3
3
f
ac
to
r
s
to
d
eter
m
in
e
th
e
f
i
n
al
d
ataset.
T
h
e
K
-
m
ea
n
s
an
d
FC
M
clu
s
ter
in
g
tech
n
iq
u
es
wer
e
th
en
u
s
ed
to
p
r
o
ce
s
s
th
e
d
ata
in
o
r
d
er
to
f
i
n
d
g
r
o
u
p
in
g
s
o
f
cr
iter
ia
with
c
o
m
p
ar
ab
le
ac
ad
em
ic
a
n
d
n
o
n
-
ac
ad
em
ic
tr
aits
th
at
m
ig
h
t
h
a
v
e
a
n
im
p
ac
t
o
n
s
t
u
d
en
t
g
r
ad
u
atio
n
.
B
y
c
o
n
v
er
tin
g
in
p
u
t
d
ata
in
to
f
u
zz
y
m
em
b
er
s
h
ip
d
eg
r
ee
s
,
ass
es
s
in
g
th
ese
v
alu
es
u
s
in
g
a
s
et
o
f
f
u
zz
y
r
u
les,
an
d
ap
p
l
y
in
g
d
ef
u
zz
if
icatio
n
to
o
b
tain
th
e
f
in
al
clu
s
ter
in
g
r
esu
lt,
f
u
zz
y
in
f
er
e
n
ce
f
ac
ilit
ates
f
lex
ib
le
d
ec
is
io
n
-
m
ak
in
g
.
Me
m
b
er
s
h
ip
v
alu
es
in
th
e
a
m
b
ig
u
o
u
s
r
an
g
e
o
f
0
.
5
0
2
3
to
0
.
5
9
0
3
in
d
icate
th
at
th
is
m
eth
o
d
ef
f
ec
ti
v
ely
co
n
tr
o
ls
u
n
ce
r
tain
ty
f
o
r
s
tu
d
en
ts
wh
o
s
e
d
ata
p
lace
m
en
ts
ar
e
n
ea
r
t
h
e
clu
s
ter
b
o
u
n
d
ar
y
.
2
.
2
.
Clus
t
er
o
ptim
iza
t
io
n us
ing
elbo
w
m
et
ho
d
E
s
tab
lis
h
in
g
th
e
ap
p
r
o
p
r
iate
clu
s
ter
co
n
f
ig
u
r
atio
n
is
ess
en
tial
in
clu
s
ter
in
g
r
esear
ch
,
as
it
d
ir
ec
tly
in
f
lu
en
ce
s
clu
s
ter
in
ter
p
r
etab
il
ity
an
d
m
o
d
el
r
eliab
ilit
y
[
2
8
]
.
A
m
o
d
el
th
at
o
v
e
r
f
its
th
e
d
ata
o
r
h
as p
o
o
r
clu
s
ter
s
ep
ar
atio
n
ca
n
r
esu
lt f
r
o
m
s
elec
tin
g
an
in
co
r
r
ec
t v
alu
e
o
f
k
,
t
h
er
eb
y
l
o
wer
in
g
th
e
ac
c
u
r
ac
y
an
d
in
ter
p
r
etab
ilit
y
o
f
th
e
s
tu
d
y
'
s
f
in
d
in
g
s
.
T
o
d
e
ter
m
in
e
th
e
b
est
clu
s
ter
s
tr
u
ctu
r
e,
th
is
s
tu
d
y
u
s
ed
th
e
e
l
b
o
w
m
eth
o
d
,
with
th
e
s
u
m
o
f
s
q
u
ar
e
d
er
r
o
r
s
(
SS
E
)
a
s
th
e
b
en
ch
m
a
r
k
.
SS
E
c
o
m
p
u
t
es
t
h
e
a
g
g
r
e
g
at
e
s
q
u
a
r
e
d
d
is
t
an
ce
b
etw
ee
n
ea
c
h
d
a
ta
p
o
i
n
t
a
n
d
t
h
e
wei
g
h
t
ed
ce
n
t
r
o
id
i
n
FC
M
o
r
t
h
e
cl
u
s
t
er
c
e
n
t
r
o
id
ass
ig
n
ed
t
o
th
e
d
ata
p
o
in
t
i
n
K
-
m
e
a
n
s
.
Sta
te
d
d
i
f
f
e
r
e
n
t
ly
,
SS
E
q
u
an
ti
f
ies
t
h
e
d
e
g
r
ee
t
o
w
h
ic
h
t
h
e
d
at
a
p
o
in
ts
c
lu
s
te
r
ar
o
u
n
d
t
h
e
ce
n
t
e
r
o
f
ea
ch
i
n
d
i
v
i
d
u
al
c
lu
s
te
r
.
SS
E
ca
n
b
e
s
tat
e
d
m
at
h
e
m
a
tic
all
y
as
(
1
)
.
W
h
e
r
e
x
is
t
h
e
d
a
ta
p
o
i
n
t
,
C
i
r
e
p
r
es
en
t
s
t
h
e
i
th
cl
u
s
te
r
,
a
n
d
o
f
t
h
e
ce
n
tr
o
i
d
o
f
cl
u
s
te
r
i
.
=
∑
∑
|
|
−
|
|
2
=
1
(
1
)
Fo
r
K
-
m
ea
n
s
,
SS
E
o
n
ly
s
h
o
ws
h
o
w
clo
s
e
to
g
eth
er
g
r
o
u
p
s
ar
e
ar
o
u
n
d
clea
r
ce
n
te
r
s
[
7
]
,
[
2
9
]
–
[
3
2
]
.
Fu
zz
y
SS
E
i
s
m
o
r
e
s
en
s
itiv
e
in
id
en
tify
in
g
o
v
e
r
lap
p
in
g
d
at
a
g
r
o
u
p
s
in
FC
M
s
in
ce
it
tak
es
in
to
ac
co
u
n
t
th
e
m
em
b
er
s
h
ip
weig
h
ts
o
f
ea
ch
d
ata
p
o
in
t.
I
n
th
e
m
ea
n
tim
e,
th
e
e
lb
o
w
m
eth
o
d
is
em
p
lo
y
ed
to
ascer
tain
th
e
p
r
o
p
er
v
alu
e
o
f
k
s
o
th
at
th
e
r
esu
ltin
g
clu
s
ter
s
ar
e
b
o
th
ea
s
ily
in
ter
p
r
etab
le
an
d
s
tatis
ti
ca
lly
s
u
p
p
o
r
ted
in
b
o
th
ap
p
r
o
ac
h
es.
T
h
e
d
eg
r
ee
o
f
m
em
b
er
s
h
ip
is
tak
en
in
to
ac
co
u
n
t
wh
ile
u
s
in
g
th
e
FC
M
ap
p
r
o
ac
h
.
B
ec
au
s
e
ea
ch
d
ata
p
o
in
t
m
ay
p
ar
tially
b
elo
n
g
to
m
o
r
e
th
an
o
n
e
clu
s
ter
,
th
e
SS
E
ca
n
b
e
m
ea
s
u
r
ed
in
a
f
u
zz
y
f
o
r
m
.
T
h
e
e
lb
o
w
m
eth
o
d
is
th
en
u
s
ed
b
y
c
o
m
p
u
tin
g
th
e
SS
E
v
alu
es
f
o
r
r
an
g
e
o
f
p
o
ten
tial
clu
s
ter
co
u
n
ts
,
f
r
o
m
k
=
2
to
k
=
1
0
.
T
h
e
SS
E
v
alu
e
will
o
f
ten
d
r
o
p
as
th
e
n
u
m
b
er
o
f
clu
s
ter
s
r
is
es.
B
u
t
th
e
d
ec
lin
e
b
ec
o
m
es
les
s
n
o
ticea
b
le
af
ter
a
ce
r
tain
p
o
in
t.
T
h
is
p
o
in
t
is
t
h
en
u
s
ed
to
b
alan
ce
er
r
o
r
r
e
d
u
ctio
n
a
n
d
m
o
d
el
co
m
p
lex
it
y
to
f
in
d
th
e
id
ea
l
n
u
m
b
er
o
f
cl
u
s
ter
s
.
T
h
e
s
ilh
o
u
ette
co
ef
f
icien
t
(
SC
)
is
u
s
ed
to
v
er
if
y
th
e
q
u
ality
o
f
th
e
clu
s
ter
in
g
r
esu
lts
af
ter
th
e
e
lb
o
w
m
eth
o
d
d
eter
m
in
es
th
e
id
ea
l
n
u
m
b
er
o
f
clu
s
ter
s
.
Stro
n
g
in
te
r
n
al
s
im
ilar
ity
an
d
d
is
tin
ct
s
ep
ar
atio
n
f
r
o
m
o
th
er
clu
s
ter
s
ar
e
g
u
ar
a
n
teed
b
y
th
i
s
v
alid
atio
n
.
T
h
e
r
ef
o
r
e,
m
o
r
e
d
ep
en
d
ab
le
an
d
ac
cu
r
ate
clu
s
ter
in
g
f
in
d
in
g
s
o
n
d
atasets
lin
k
ed
to
g
r
ad
u
atio
n
r
eq
u
ir
em
en
ts
ca
n
b
e
o
b
tain
e
d
u
s
in
g
a
two
-
s
tag
e
o
p
tim
izatio
n
p
r
o
ce
d
u
r
e
th
at
in
clu
d
es
e
lb
o
w
-
SS
E
to
s
elec
t th
e
ca
n
d
id
ate
n
u
m
b
er
o
f
clu
s
ter
s
an
d
s
ilh
o
u
ette
an
aly
s
is
f
o
r
c
lu
s
ter
v
alid
atio
n
.
2
.
3
.
Va
lid
a
t
ing
clus
t
er
s
wit
h t
he
s
ilh
o
uet
t
e
Sil
h
o
u
ette
a
n
aly
s
is
is
em
p
lo
y
ed
to
ass
ess
clu
s
ter
v
alid
ity
b
y
ev
alu
atin
g
co
h
esio
n
an
d
th
e
d
i
s
tin
ctio
n
s
g
en
er
ated
b
y
th
e
p
r
o
p
o
s
ed
m
e
th
o
d
.
T
h
is
m
eth
o
d
allo
ws
f
o
r
a
m
o
r
e
o
b
jectiv
e
ass
ess
m
en
t
o
f
clu
s
ter
f
o
r
m
atio
n
ac
cu
r
ac
y
b
y
co
m
p
ar
in
g
th
e
e
x
ten
t
o
f
r
esem
b
lan
ce
am
o
n
g
d
ata
with
in
th
e
id
en
tical
clu
s
ter
an
d
th
e
d
eg
r
ee
o
f
d
if
f
er
en
ce
b
etwe
en
clu
s
ter
s
[
3
3
]
–
[
3
6
]
.
T
h
is
f
o
r
m
u
la
is
d
ef
in
ed
in
(
2
)
[
3
7
]
.
(
)
=
(
)
−
(
)
m
ax
{
(
)
,
(
)
}
(
2
)
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J Ar
tif
I
n
tell
I
SS
N:
2252
-
8
9
3
8
C
o
mp
a
r
is
o
n
a
lg
o
r
ith
ms w
ith
o
p
timiz
a
tio
n
fo
r
clu
s
teri
n
g
mu
lti
-
crit
eria
(
I
d
a
Mu
lya
d
i
)
3479
W
h
er
e
(
)
d
en
o
tes
th
e
lo
west
a
v
er
ag
e
d
is
tan
ce
b
etwe
en
p
o
i
n
t
i
an
d
o
b
jects
in
th
e
cl
o
s
est
n
eig
h
b
o
r
in
g
clu
s
ter
;
an
d
(
)
r
ep
r
esen
ts
th
e
av
er
ag
e
d
is
tan
ce
b
etwe
en
p
o
i
n
t
i
an
d
all
o
th
er
o
b
jects
in
th
e
s
a
m
e
clu
s
ter
.
T
h
e
r
an
g
e
o
f
v
alu
es
f
o
r
th
e
SC
is
-
1
to
+1
.
A
v
al
u
e
ar
o
u
n
d
0
in
d
icate
s
th
at
th
e
d
ata
p
o
in
t
is
cl
o
s
e
to
th
e
b
o
u
n
d
ar
y
am
o
n
g
two
clu
s
ter
s
;
a
v
alu
e
n
ea
r
+1
s
h
o
ws
th
at
th
e
p
o
in
t
h
a
s
b
ee
n
co
r
r
ec
tly
clu
s
ter
ed
in
s
i
d
e
its
clu
s
ter
,
an
d
a
v
alu
e
n
ea
r
-
1
in
d
icate
s
th
at
th
e
p
o
in
t is p
r
o
b
ab
ly
in
th
e
wr
o
n
g
clu
s
ter
.
T
h
e
o
v
er
all
SC
is
th
e
m
ea
n
o
f
all
th
e
d
ata
p
o
in
ts
.
A
clu
s
ter
with
h
ig
h
er
n
u
m
b
e
r
s
is
clo
s
e
r
to
g
eth
er
an
d
less
s
p
r
ea
d
o
u
t.
Af
ter
K
-
m
ea
n
s
an
d
FC
M
clu
s
ter
in
g
o
n
m
u
ltip
le
f
ac
to
r
s
(
GPA,
co
u
r
s
e
co
m
p
letio
n
,
cr
ed
it
lo
ad
,
atten
d
an
ce
,
ex
tr
ac
u
r
r
icu
l
ar
ac
tiv
ities
,
s
o
cio
ec
o
n
o
m
ic
b
ac
k
g
r
o
u
n
d
,
ag
e,
an
d
g
e
n
d
er
)
,
SC
was
u
s
ed
in
th
is
s
tu
d
y
.
Fo
r
K
-
m
ea
n
s
,
SC
is
u
s
ed
to
ch
ec
k
wh
eth
er
t
h
e
s
tu
d
e
n
t
g
r
o
u
p
in
g
is
co
r
r
ec
t,
wh
ile
i
n
FC
M,
S
C
is
u
s
ed
to
s
ee
f
u
zz
y
m
em
b
er
s
h
ip
.
W
h
en
u
s
ed
with
e
l
b
o
w
–
SS
E
,
S
C
g
u
ar
an
tees
ac
cu
r
ate
a
n
d
s
i
m
p
le
g
r
o
u
p
in
g
f
o
r
s
tu
d
y
in
g
f
ac
t
o
r
s
th
at
af
f
ec
t
g
r
a
d
u
atio
n
.
2
.
4
.
Clus
t
er
ing
a
lg
o
rit
h
m
s
T
h
e
s
tu
d
y
ex
am
in
e
d
d
ata
s
ets
f
r
o
m
p
r
ev
io
u
s
r
esear
ch
o
n
m
u
lti
-
cr
iter
ia,
wh
ich
co
n
tain
b
o
th
ac
ad
em
ic
an
d
n
o
n
-
ac
ad
em
ic
elem
en
ts
th
at
in
f
lu
en
ce
a
s
tu
d
en
t'
s
ca
p
a
city
to
g
r
ad
u
ate.
Usi
n
g
d
if
f
er
en
t
ty
p
es
o
f
an
aly
s
is
in
d
if
f
er
e
n
t
s
tu
d
ies
s
u
p
p
o
r
ts
th
e
id
ea
th
at
u
s
in
g
b
o
t
h
ac
ad
em
ic
an
d
n
o
n
-
ac
a
d
em
ic
f
ac
to
r
s
to
g
eth
er
m
a
k
es
g
r
ad
u
atio
n
e
x
p
ec
tatio
n
s
m
o
r
e
ac
cu
r
ate
th
an
u
s
in
g
o
n
ly
o
n
e
ty
p
e
o
f
f
ac
to
r
.
Stra
teg
ies
t
o
r
aise
th
e
g
r
a
d
u
atio
n
r
ate
o
f
s
tu
d
en
ts
d
o
n
'
t
ju
s
t
f
o
c
u
s
o
n
im
p
r
o
v
in
g
ac
a
d
em
ic
p
e
r
f
o
r
m
a
n
ce
;
th
e
y
also
lo
o
k
at
th
in
g
s
lik
e
s
tu
d
en
ts
'
d
r
iv
e,
th
e
s
u
p
p
o
r
t
th
ey
g
et
in
th
e
class
r
o
o
m
,
an
d
th
eir
s
o
cial
s
itu
atio
n
s
to
m
ak
e
a
m
o
r
e
co
m
p
lete
an
d
lo
n
g
-
last
in
g
p
la
n
[
3
8
]
.
2
.
4
.
1
.
Clus
t
er
ing
us
i
ng
K
-
m
e
a
ns
f
o
r
m
ultiple c
rit
er
ia
Gr
ad
u
atio
n
-
r
elate
d
cr
iter
ia
a
r
e
g
r
o
u
p
e
d
in
t
o
ac
a
d
em
ic
a
n
d
n
o
n
-
ac
ad
em
ic
ca
te
g
o
r
ies
u
s
in
g
th
e
K
-
m
ea
n
s
clu
s
ter
in
g
ap
p
r
o
ac
h
.
T
h
is
tech
n
iq
u
e
p
ar
titi
o
n
s
th
e
d
ata
in
to
a
f
ix
ed
n
u
m
b
er
o
f
cl
u
s
ter
s
,
ea
ch
with
a
ce
n
tr
o
id
.
T
o
lim
it
v
ar
ia
n
ce
with
in
th
e
clu
s
ter
u
n
til
a
co
n
v
er
g
en
ce
co
n
d
itio
n
is
m
et,
t
h
e
cl
u
s
ter
m
em
b
er
s
h
ip
is
u
p
d
ated
b
y
f
ir
s
t
co
m
p
u
tin
g
t
h
e
E
u
clid
ea
n
d
is
tan
ce
f
r
o
m
ea
ch
d
ata
p
o
in
t
to
th
e
ce
n
tr
o
id
[
3
9
]
–
[
4
5
]
.
T
h
is
ap
p
r
o
ac
h
p
a
r
titi
o
n
s
th
e
d
ata
in
to
a
p
r
ed
eter
m
in
ed
q
u
a
n
tity
.
o
f
clu
s
ter
s
,
.
A
ce
n
t
r
al
p
o
in
t,
k
n
o
wn
as
t
h
e
ce
n
tr
o
id
,
is
th
en
u
s
ed
t
o
r
ep
r
e
s
en
t
ea
ch
clu
s
ter
.
T
h
e
K
-
m
ea
n
s
alg
o
r
ith
m
iter
ativ
ely
ass
ig
n
s
o
b
s
er
v
atio
n
s
to
th
e
clo
s
est ce
n
tr
o
id
an
d
u
p
d
ates th
e
ce
n
tr
o
id
p
o
s
itio
n
s
u
n
til a
s
t
ab
le
p
ar
titi
o
n
is
attain
ed
.
T
h
e
ca
lcu
latio
n
ty
p
ically
u
s
es
E
u
clid
ea
n
d
is
tan
ce
as sh
o
wn
in
(
3
)
.
(
,
)
=
√
(
1
−
1
)
2
+
(
2
−
2
)
2
+
⋯
+
(
−
)
2
(
3
)
W
h
er
e
=
(
1
,
2
,
…
,
)
is
d
ata
p
o
in
t
an
d
=
(
1
,
2
,
…
,
)
is
th
e
ce
n
tr
o
id
,
u
p
d
ate
ce
n
tr
o
id
in
(
4
)
.
=
1
|
|
∑
(
4
)
W
h
er
e
is
a
co
llectio
n
o
f
d
ata
in
clu
s
ter
I
,
|
|
is
th
e
n
u
m
b
e
r
o
f
d
ata
in
th
e
clu
s
ter
.
‒
C
o
n
v
er
g
en
ce
:
u
n
til
t
h
e
m
a
x
im
u
m
iter
atio
n
lim
it
is
r
ea
c
h
e
d
o
r
th
e
ce
n
tr
o
i
d
ch
a
n
g
es
o
n
l
y
s
lig
h
tly
,
t
h
is
p
r
o
ce
s
s
is
r
ep
ea
ted
.
‒
R
esu
lts
:
f
in
al
ce
n
tr
o
id
:
th
e
ce
n
tr
o
id
’
s
u
ltima
te
lo
ca
tio
n
f
o
r
ev
er
y
clu
s
ter
.
T
h
e
clu
s
ter
lab
el
is
b
ased
o
n
h
o
w
clo
s
e
it is
to
th
e
n
ea
r
est c
lu
s
ter
; e
ac
h
d
ata
p
o
i
n
t is lab
eled
ac
co
r
d
i
n
g
ly
.
2
.
4
.
2
.
Clus
t
er
ing
us
i
ng
f
uzzy
C
-
m
ea
ns
f
o
r
m
ultiple c
rit
er
ia
T
h
e
m
eth
o
d
k
n
o
wn
as
FC
M
was
u
s
ed
in
th
is
s
tu
d
y
to
g
r
o
u
p
s
tu
d
en
ts
t
o
g
eth
er
b
ased
o
n
ac
ad
em
ic
an
d
n
o
n
-
ac
a
d
em
ic
f
ac
to
r
s
th
at
af
f
ec
t
th
eir
c
h
an
ce
s
o
f
g
r
a
d
u
atin
g
[
4
6
]
.
B
y
u
s
in
g
th
e
s
o
f
t
c
lu
s
ter
in
g
ap
p
r
o
ac
h
,
FC
M
a
s
s
ig
n
s
a
m
em
b
er
s
h
ip
lev
el
to
o
n
e
o
r
m
o
r
e
clu
s
ter
s
in
ad
d
itio
n
to
p
lacin
g
ea
ch
cr
i
ter
io
n
ab
s
o
lu
tely
in
o
n
e
clu
s
ter
[
4
7
]
–
[
4
9
]
.
Ho
w
th
e
alg
o
r
ith
m
c
o
n
tin
u
o
u
s
ly
m
o
d
if
ies ce
n
tr
o
id
s
an
d
m
e
m
b
er
s
h
i
p
v
alu
es:
i)
Set u
p
th
e
m
em
b
er
s
h
ip
m
atr
ix
(
)
=
[
u
ij
]
r
an
d
o
m
ly
,
e
n
s
u
r
in
g
∑
=
1
=
1
ii)
C
o
m
p
u
te
clu
s
ter
ce
n
tr
o
i
d
s
in
(
5
)
.
=
∑
.
=
1
∑
=
1
(
5
)
iii)
Up
d
ate
m
em
b
er
s
h
i
p
v
alu
es
ar
e
u
p
d
ated
ac
co
r
d
i
n
g
to
(
6
)
.
=
1
∑
(
|
|
−
|
|
|
|
−
|
|
)
2
−
1
=
1
(
6
)
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8
9
3
8
I
n
t J Ar
tif
I
n
tell
,
Vo
l.
15
,
No
.
4
,
Au
g
u
s
t
20
26
:
3
4
7
6
-
3
4
9
1
3480
iv
)
T
h
e
o
b
jectiv
e
f
u
n
ctio
n
to
b
e
m
in
im
ized
is
g
iv
en
in
(
7
)
.
(
)
=
∑
∑
(
)
.
2
(
,
(
)
)
=
1
=
1
(
7
)
v)
R
ep
ea
t
s
tep
s
ii
–
iii
ag
ain
an
d
a
g
ain
u
n
til
co
n
v
er
g
e
n
ce
,
wh
ic
h
m
ea
n
s
th
at
t
h
e
ch
a
n
g
e
in
is
less
th
an
th
e
k
ey
v
alu
e
.
GPA
co
u
r
s
e
co
m
p
letio
n
r
ate
,
n
u
m
b
er
o
f
r
ep
ea
te
d
co
u
r
s
e
s
,
cr
ed
it
lo
a
d
,
atten
d
an
ce
r
a
te,
ex
tr
ac
u
r
r
icu
lar
in
v
o
lv
em
e
n
t,
s
o
cio
ec
o
n
o
m
ic
b
ac
k
g
r
o
u
n
d
,
ag
e,
an
d
g
en
d
er
ar
e
s
o
m
e
o
f
th
e
f
ac
to
r
s
th
at
wer
e
u
s
ed
in
th
is
s
tu
d
y
.
all
v
alu
es
ar
e
ad
ju
s
ted
t
o
f
all
b
etwe
en
0
an
d
1
u
s
in
g
th
e
m
in
-
m
a
x
s
ca
le.
2
.
5
.
O
pti
m
iza
t
io
n us
ing
pa
rt
icle
s
wa
rm
o
ptim
iza
t
i
o
n a
nd
g
enet
ic
a
lg
o
rit
h
m
C
lu
s
ter
in
g
o
p
tim
izatio
n
ca
n
b
e
ca
r
r
ied
o
u
t
u
s
in
g
GA
an
d
PS
O
to
im
p
r
o
v
e
th
e
q
u
ality
o
f
clu
s
ter
in
g
f
in
d
in
g
s
.
W
h
ile
PS
O
m
o
d
if
ies
ce
n
tr
o
id
p
lace
m
en
ts
b
ased
o
n
th
e
b
est
p
a
r
ticle'
s
ex
p
er
ien
ce
an
d
th
e
b
est
g
lo
b
al
s
o
lu
tio
n
,
GA
c
r
ea
tes
th
e
o
p
tim
al
ce
n
tr
o
i
d
c
o
n
f
ig
u
r
atio
n
th
r
o
u
g
h
s
elec
tio
n
,
cr
o
s
s
o
v
er
,
a
n
d
m
u
tatio
n
o
p
er
atio
n
s
.
T
h
e
o
b
jectiv
e
f
u
n
c
tio
n
to
b
e
m
in
im
ized
is
eith
er
J
m
in
FC
M
o
r
in
er
tia
in
K
-
m
e
an
s
.
B
y
u
s
in
g
b
o
th
s
tr
ateg
ies,
th
is
m
eth
o
d
ca
n
p
r
e
v
en
t
lo
ca
l
m
in
im
a,
th
e
PS
O
an
d
GA
im
p
r
o
v
e
clu
s
ter
c
o
h
esio
n
an
d
d
is
s
o
ciatio
n
,
an
d
r
esu
lt in
m
o
r
e
ef
f
ec
tiv
e
cl
u
s
ter
in
g
p
r
o
ce
d
u
r
e
o
n
co
m
p
lic
ated
d
atasets
.
T
h
e
f
o
r
m
u
las ar
e
as
in
(
8
)
a
n
d
(
9
)
.
(
+
1
)
=
.
(
)
+
1
1
(
−
(
)
)
+
2
2
(
−
(
)
)
(
+
1
)
=
(
)
+
(
+
1
)
(
8
)
=
1
(
)
(
−
)
(
9
)
3.
RE
SU
L
T
S AN
D
D
I
SCU
SS
I
O
N
3
.
1
.
Da
t
a
s
et
T
ab
le
1
s
h
o
ws
th
at
th
e
d
ata
u
s
ed
in
th
is
s
tu
d
y
co
n
s
is
ts
o
f
3
3
cr
iter
ia
an
d
1
6
e
x
p
er
ts
wh
o
p
r
o
v
id
e
d
ass
es
s
m
en
ts
f
o
r
ea
ch
cr
iter
io
n
.
W
eig
h
tin
g
is
g
iv
en
b
as
ed
o
n
th
e
lev
el
o
f
f
ac
to
r
s
in
f
lu
en
cin
g
s
tu
d
e
n
t
g
r
ad
u
atio
n
,
s
u
c
h
as
weig
h
ts
0
.
4
3
7
9
6
(
v
er
y
in
f
lu
en
tial)
,
0
.
2
1
8
9
8
(
in
f
lu
en
tial)
,
an
d
0
.
1
4
5
9
9
(
less
in
f
lu
e
n
tial)
.
T
h
ese
d
if
f
er
en
ce
s
in
w
h
at
ex
p
er
ts
ag
r
ee
o
n
a
r
e
im
p
o
r
tan
t
in
m
u
lti
-
cr
iter
ia
d
ec
is
io
n
-
m
a
k
in
g
b
ec
au
s
e
th
ey
a
d
d
v
alu
e
to
th
e
r
ev
iew
an
d
h
elp
u
s
u
n
d
er
s
tan
d
wh
y
p
eo
p
le
h
av
e
d
if
f
er
en
t
id
ea
s
.
B
y
u
n
itin
g
th
e
v
iews
o
f
th
e
m
ajo
r
ity
an
d
m
in
o
r
ity
,
th
e
d
e
cisi
o
n
will
b
e
m
o
r
e
co
m
p
lete,
s
tr
o
n
g
er
,
an
d
o
p
en
to
ev
er
y
o
n
e.
T
ab
le
2
d
is
p
lay
s
th
e
o
u
tco
m
es
o
f
ap
p
ly
in
g
h
ig
h
-
d
im
en
s
io
n
al
h
a
n
d
lin
g
with
an
a
u
to
en
co
d
er
p
r
i
o
r
to
clu
s
ter
in
g
,
u
tili
zin
g
two
m
eth
o
d
s
:
en
co
d
er
(
co
m
p
r
ess
ed
f
ea
tu
r
es)
an
d
d
ec
o
d
er
(
r
ec
o
n
s
tr
u
ctin
g
th
e
o
r
ig
in
al
d
ata)
,
wh
ich
r
ed
u
ce
s
th
e
n
u
m
b
er
o
f
c
r
iter
ia
f
r
o
m
3
3
,
1
6
f
ea
tu
r
es to
2
f
ea
tu
r
es.
T
ab
le
1
.
Data
s
et
o
f
ass
ess
m
en
ts
f
o
r
ea
ch
cr
iter
io
n
Ex
p
e
r
t
C
r
i
t
1
C
r
i
t
2
C
r
i
t
3
C
r
i
t
4
C
r
i
t
5
C
r
i
t
6
C
r
i
t
7
C
r
i
t
8
.
.
C
r
i
t
3
3
1
0
.
4
3
7
9
6
0
.
2
1
8
9
8
0
.
4
3
7
9
6
0
.
4
3
7
9
6
0
.
4
3
7
9
6
0
.
2
1
8
9
8
0
.
4
3
7
9
6
0
.
4
3
7
9
6
0
.
4
3
7
9
6
2
0
.
4
3
7
9
6
0
.
2
1
8
9
8
0
.
4
3
7
9
6
0
.
4
3
7
9
6
0
.
4
3
7
9
6
0
.
2
1
8
9
8
0
.
4
3
7
9
6
0
.
4
3
7
9
6
0
.
4
3
7
9
6
3
0
.
4
3
7
9
6
0
.
4
3
7
9
6
0
.
2
1
8
9
8
0
.
2
1
8
9
8
0
.
4
3
7
9
6
0
.
4
3
7
9
6
0
.
4
3
7
9
6
0
.
4
3
7
9
6
0
.
4
3
7
9
6
4
0
.
4
3
7
9
6
0
.
4
3
7
9
6
0
.
4
3
7
9
6
0
.
4
3
7
9
6
0
.
4
3
7
9
6
0
.
4
3
7
9
6
0
.
2
1
8
9
8
0
.
4
3
7
9
6
0
.
4
3
7
9
6
5
.
.
0
.
4
3
7
9
6
0
.
4
3
7
9
6
0
.
4
3
7
9
6
0
.
4
3
7
9
6
0
.
2
1
8
9
8
0
.
4
3
7
9
6
0
.
2
1
8
9
8
0
.
2
1
8
9
8
0
.
4
3
7
9
6
16
0
.
2
1
8
9
8
0
.
4
3
7
9
6
0
.
4
3
7
9
6
0
.
2
1
8
9
8
0
.
4
3
7
9
6
0
.
2
1
8
9
8
0
.
2
1
8
9
8
0
.
2
1
8
9
8
0
.
2
1
8
9
8
T
ab
le
2
.
Featu
r
e
c
o
m
p
r
ess
io
n
r
esu
lts
with
au
to
en
co
d
e
r
N
u
mb
e
r
0
1
2
3
.
.
32
La
t
e
n
t
_
F
e
a
t
u
r
e
_
1
0
.
1
1
2
3
8
1
0
.
1
1
2
3
8
1
0
.
1
1
2
3
8
1
0
.
1
1
2
3
8
1
-
2
.
8
2
7
4
4
9
La
t
e
n
t
_
F
e
a
t
u
r
e
_
2
0
.
8
1
6
4
5
2
0
.
8
1
6
4
5
2
0
.
8
1
6
4
5
2
0
.
8
1
6
4
5
2
-
2
.
7
2
1
4
5
8
C
l
u
st
e
r
0
0
0
0
1
3
.
2
.
E
lbo
w
wit
h
s
ilh
o
uet
t
e
A
m
o
r
e
o
b
jectiv
e
f
o
u
n
d
atio
n
f
o
r
f
ig
u
r
in
g
o
u
t
th
e
id
ea
l
n
u
m
b
er
o
f
clu
s
ter
s
ca
n
b
e
o
b
tain
ed
b
y
co
m
b
in
in
g
th
e
e
lb
o
w
tech
n
iq
u
e
with
th
e
s
ilh
o
u
ette
s
co
r
e.
T
h
is
co
m
b
in
atio
n
e
v
alu
ates
th
e
q
u
ality
o
f
clu
s
ter
cr
ea
tio
n
b
y
co
n
s
id
er
i
n
g
th
e
d
eg
r
ee
o
f
s
ep
ar
atio
n
b
etwe
en
clu
s
ter
s
an
d
th
e
p
o
ten
tial
f
o
r
o
v
er
lap
.
T
h
e
e
lb
o
w
tech
n
iq
u
e
u
s
es th
e
p
atter
n
o
f
SS
E
r
ed
u
ctio
n
t
o
ascer
tain
th
e
o
p
tim
al
q
u
an
tity
o
f
clu
s
ter
s
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J Ar
tif
I
n
tell
I
SS
N:
2252
-
8
9
3
8
C
o
mp
a
r
is
o
n
a
lg
o
r
ith
ms w
ith
o
p
timiz
a
tio
n
fo
r
clu
s
teri
n
g
mu
lti
-
crit
eria
(
I
d
a
Mu
lya
d
i
)
3481
3
.
2
.
1
.
E
lbo
w
T
ab
le
2
d
is
p
lay
s
th
e
o
u
tco
m
es
o
f
th
e
SS
E
co
m
p
u
tatio
n
u
s
in
g
(
1
)
.
E
v
e
r
y
d
ata
p
o
i
n
t
(
)
is
d
eter
m
in
ed
,
g
r
o
u
p
e
d
ac
c
o
r
d
in
g
to
th
e
ce
n
tr
o
id
(
)
,
an
d
th
e
s
q
u
ar
ed
d
is
tan
ce
b
etwe
en
ea
ch
d
a
ta
p
o
in
t
an
d
th
e
ass
o
ciate
d
ce
n
tr
o
id
(
|
|
−
|
|
2
)
is
th
en
ca
lcu
lated
.
T
h
e
SS
E
is
th
en
d
eter
m
in
ed
b
y
s
u
m
m
in
g
th
e
s
q
u
ar
e
d
d
is
tan
ce
s
o
f
ea
c
h
p
o
in
t
with
in
ea
ch
clu
s
ter
.
On
e
o
f
th
e
co
m
p
u
tatio
n
o
u
tp
u
ts
,
f
o
r
in
s
tan
ce
,
p
r
o
d
u
ce
d
an
SS
E
v
alu
e
o
f
3
7
1
.
9
1
1
8
9
8
.
I
n
th
e
f
o
llo
win
g
r
o
u
n
d
,
th
is
v
alu
e
d
r
o
p
p
e
d
d
r
am
atica
lly
u
n
til
it
r
ea
ch
ed
an
SS
E
o
f
3
2
3
.
7
7
1
5
7
7
at
k
=3
.
T
h
is
d
ec
li
n
e
s
u
g
g
ests
th
at
th
e
b
est clu
s
ter
in
g
o
u
tco
m
e
is
th
r
ee
clu
s
ter
s
[
5
0
]
.
=
[
0
.
43796
,
0
.
43796
,
0
.
43796
,
0
.
43796
,
0
.
43796
,
0
.
43796
,
0
.
43796
,
0
.
43796
,
0
.
43796
,
0
.
43796
,
0
.
43796
,
…
]
=
[
0
.
3384
,
0
.
3008
,
0
.
3296
,
0
.
3119
,
0
.
3097
,
0
.
3362
,
0
.
2964
,
0
.
3008
,
0
.
2853
,
0
.
2853
,
0
.
2920
,
0
.
2676
,
0
.
3163
,
.
.
]
|
|
−
|
|
=
[
(
0
.
43796
−
0
.
3384
)
2
+
(
43796
−
0
.
3008
)
2
+
(
0
.
43796
−
0
.
3296
)
2
+
(
0
.
43796
−
0
.
3119
)
2
+
⋯
]
=
371
.
911898
3
.
2
.
2
.
Sil
ho
uet
t
e
Sil
h
o
u
ette
is
th
en
u
s
ed
to
p
r
o
ce
s
s
th
e
elb
o
w's
k
=3
v
alu
e.
Fo
r
ev
er
y
d
ata
p
o
in
t
,
two
v
alu
es
ar
e
co
m
p
u
ted
:
(
)
,
wh
ich
is
th
e
av
e
r
ag
e
d
is
tan
ce
to
th
e
clo
s
est
o
th
er
clu
s
ter
u
s
ed
to
m
ea
s
u
r
e
s
ep
ar
atio
n
,
a
n
d
(
)
[
5
1
]
–
[
5
3
]
.
I
n
(
2
)
is
u
s
ed
to
ca
lcu
late
th
e
s
ilh
o
u
ette
v
alu
e,
wh
ich
r
an
g
es
f
r
o
m
−1
to
1
.
T
h
e
clu
s
ter
'
s
q
u
ality
is
ev
alu
ated
u
s
in
g
t
h
e
s
ilh
o
u
ette
s
co
r
e
,
wh
er
e
a
n
u
m
b
er
n
ea
r
1
d
en
o
tes
a
g
o
o
d
cl
u
s
ter
,
a
v
alu
e
n
ea
r
0
in
d
icate
s
th
at
th
e
d
ata
is
clo
s
e
to
th
e
clu
s
ter
b
o
u
n
d
ar
y
,
an
d
a
n
eg
ativ
e
v
alu
e
s
u
g
g
ests
a
p
o
ten
tial
clu
s
ter
in
g
er
r
o
r
.
T
h
e
to
tal
s
ilh
o
u
ette
v
al
u
e
is
ca
lcu
lated
b
y
av
er
ag
i
n
g
ea
ch
d
ata
s
co
r
e.
Fin
d
in
g
t
h
e
av
er
ag
e
d
is
tan
ce
b
etwe
en
m
em
b
er
s
o
f
ea
ch
cl
u
s
ter
is
th
e
f
ir
s
t
s
tag
e
in
th
i
s
co
m
p
u
tatio
n
.
Fo
r
i
n
s
tan
ce
,
clu
s
ter
0
h
as
ten
m
em
b
er
s
,
wh
ile
clu
s
ter
1
h
as twe
n
ty
-
th
r
ee
.
C
o
m
p
u
tin
g
t
h
e
d
i
s
tan
ce
s
(
)
,
(
)
,
an
d
(
)
.
(
)
=
√
(
0
.
43796
−
0
.
21898
)
2
+
⋯
+
(
0
.
43796
−
0
.
21898
)
2
=
3
.
574779
(
)
=
√
(
0
.
43796
−
0
.
21898
)
2
+
⋯
+
(
0
.
43796
−
0
.
21898
)
2
=6
.
8
0
3
6
6
6
ℎ
=
6
.
803666
−
3
.
574779
6
.
803666
=
0
.
266789
T
ab
le
3
s
h
o
ws
th
e
SS
E
v
alu
e
d
r
o
p
p
e
d
f
r
o
m
3
2
3
.
9
1
1
9
at
k
=
3
to
1
7
3
.
1
5
2
9
1
at
k
=
1
0
,
s
u
g
g
esti
n
g
th
at
as
th
e
n
u
m
b
er
o
f
clu
s
ter
s
r
is
es,
th
e
d
is
tan
ce
b
etwe
en
d
ata
p
o
in
ts
an
d
ce
n
tr
o
id
s
d
ec
r
e
ases
.
B
u
t
th
e
SS
E
v
alu
e
is
n
o
t
th
e
o
n
ly
f
ac
to
r
th
at
d
eter
m
in
es
th
e
o
p
tim
al
n
u
m
b
er
o
f
clu
s
ter
s
.
T
h
e
si
lh
o
u
et
te
s
co
r
e
r
ea
ch
ed
its
m
ax
im
u
m
v
alu
e
o
f
0
.
2
6
6
7
8
9
at
k
=
2
.
Fr
o
m
k
=
3
to
k
=
7
,
th
e
s
co
r
e
d
r
o
p
p
e
d
;
it
th
en
in
cr
ea
s
ed
s
lig
h
tly
at
k
=
8
an
d
k
=
9
,
b
u
t
r
em
ain
e
d
b
elo
w
th
e
p
ea
k
v
alu
e
at
k
=
2
.
T
h
is
d
em
o
n
s
tr
ates
th
at
th
e
o
p
tim
al
n
u
m
b
er
o
f
clu
s
ter
s
f
o
r
th
is
d
ataset
is
k
=
2.
W
ith
th
e
elb
o
w
p
o
in
t
ap
p
ar
en
t
at
k
=
3
,
th
e
elb
o
w
tech
n
iq
u
e
d
em
o
n
s
tr
ates
th
at
th
e
SS
E
v
alu
e
f
alls
as
th
e
n
u
m
b
er
o
f
clu
s
ter
s
r
is
es.
Sin
c
e
th
e
s
ilh
o
u
ette
s
co
r
e
r
ea
c
h
es
its
g
r
ea
test
v
alu
e
at
k
=
2
,
k
=
2
is
th
e
i
d
ea
l n
u
m
b
e
r
o
f
clu
s
ter
s
f
o
r
th
is
d
ataset.
T
ab
le
3
.
R
esu
lt o
f
elb
o
w
an
d
s
ilh
o
u
ette
av
er
ag
e
k
I
n
e
r
t
i
a
(
S
S
E)
S
i
l
h
o
u
e
t
t
e
sc
o
r
e
2
3
7
1
.
9
1
1
9
0
.
2
6
6
7
8
9
3
3
2
3
.
7
7
1
5
8
0
.
1
9
6
3
4
1
4
2
9
0
.
6
5
4
9
2
0
.
1
8
9
4
5
5
5
..
2
6
5
.
1
0
6
4
8
0
.
1
8
3
1
5
5
10
1
7
3
.
1
5
2
9
1
0
.
1
5
2
5
4
9
3
.
3
.
Clus
t
er
a
lg
o
ri
t
hm
s
T
h
is
s
tu
d
y
em
p
lo
y
s
an
u
n
lab
el
ed
d
ata
clu
s
ter
in
g
tech
n
iq
u
e
to
clu
s
ter
a
m
u
lti
-
cr
iter
ia
d
ataset
b
ased
o
n
s
im
ilar
ity
p
atter
n
s
.
I
t
also
co
m
p
ar
es
s
ev
er
al
clu
s
ter
in
g
tech
n
iq
u
es,
in
clu
d
i
n
g
K
-
m
ea
n
s
,
FC
M,
K
-
m
ed
o
id
s
,
DB
S
C
AN,
an
d
s
p
ec
tr
al
clu
s
ter
in
g
[
5
4
]
–
[
5
6
]
.
T
h
e
co
m
p
ar
is
o
n
aim
s
to
i
d
en
tify
th
e
b
est
w
ay
to
cr
ea
te
p
r
ec
is
e
an
d
s
ig
n
if
ican
t c
lu
s
ter
s
.
3
.
3
.
1
.
K
-
m
ea
ns
T
h
e
v
alu
e
o
f
k
i
s
d
et
er
m
in
e
d
b
y
s
e
le
ct
in
g
th
e
o
p
tim
al
c
l
u
s
ter
u
s
in
g
th
e
h
y
b
r
id
e
l
b
o
w
-
s
ilh
o
u
et
te
m
eth
o
d
ap
p
l
ied
to
K
-
Me
an
s
,
as
i
llu
s
tr
a
te
d
in
Fi
g
u
r
e
2
.
T
h
e
K
-
m
e
an
s
a
lg
o
r
i
th
m
r
e
s
o
lu
t
i
o
n
p
r
o
c
e
s
s
co
n
s
i
s
t
s
o
f
f
iv
e
s
t
ag
e
s
.
T
ab
le
4
s
h
o
w
s
th
e
in
i
t
ia
l
s
ta
g
e
o
f
r
an
d
o
m
l
y
d
et
er
m
in
in
g
th
e
c
en
tr
o
id
[
5
7
]
–
[
5
9
]
,
wh
e
r
e
th
e
d
ata
i
s
t
ak
en
f
r
o
m
T
ab
le
1
.
T
ab
l
e
5
.
s
h
o
wn
E
u
cl
id
e
an
d
i
s
t
an
c
e
to
in
i
t
ial
c
en
t
r
o
id
,
t
w
o
in
it
ia
l
ce
n
tr
o
i
d
s
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8
9
3
8
I
n
t J Ar
tif
I
n
tell
,
Vo
l.
15
,
No
.
4
,
Au
g
u
s
t
20
26
:
3
4
7
6
-
3
4
9
1
3482
(
clu
s
ter
0
,
c
lu
s
te
r
1
)
ar
e
s
h
o
wn
in
r
o
w
s
(
0
,
1
)
.
On
e
o
f
th
e
d
a
ta
's
f
ea
tu
r
e
s
i
s
co
lu
m
n
(
0
.
.
.
1
5
)
.
As
a
r
e
s
u
lt
,
ev
er
y
ce
n
tr
o
id
p
o
s
s
e
s
s
e
s
m
u
l
t
id
im
en
s
io
n
a
l c
o
o
r
d
in
at
e
s
b
a
s
e
d
o
n
E
u
cl
id
e
an
d
is
ta
n
ce
in
(
1
)
.
I
n
co
m
p
a
r
is
o
n
to
C
1
,
th
e
d
is
tan
ce
b
etwe
en
all
d
ata
(
cr
iter
i
a
1
–
3
3
)
an
d
ce
n
tr
o
id
C
0
is
th
e
s
h
o
r
test
.
T
h
is
in
d
icate
s
th
at
all
o
f
t
h
e
d
ata
will
p
r
o
b
a
b
ly
b
e
p
lace
d
i
n
clu
s
ter
C
0
d
u
r
in
g
th
e
f
ir
s
t
cl
u
s
ter
in
g
cy
cle.
T
h
e
clu
s
ter
in
g
r
esu
lts
,
wh
ich
s
h
o
w
two
clu
s
ter
s
—
clu
s
ter
0
with
1
0
m
em
b
e
r
s
an
d
clu
s
ter
1
with
2
3
m
em
b
er
s
—
ar
e
s
h
o
wn
in
T
ab
le
6
.
T
h
e
m
em
b
e
r
r
eq
u
ir
e
m
en
ts
ar
e
d
is
p
lay
e
d
f
o
r
ea
ch
clu
s
ter
.
,
(
0
1
)
=
√
(
0
.
474039
−
(
−
0
.
2061
)
)
2
+
(
0
.
776439
−
(
−
0
.
33758
)
)
2
+
(
0
.
353666
−
(
−
0
.
15377
)
)
2
+
⋯
(
0
,
1
)
=
2
.
083304
Fig
u
r
e
2
.
E
l
b
o
w
an
d
s
ilh
o
u
ett
e
v
is
u
aliza
tio
n
T
ab
le
4
.
I
n
itial
ce
n
tr
o
id
C
l
u
s
t
e
r
0
1
2
3
4
5
6
7
8
.
.
15
C
l
u
s
t
e
r
0
0
.
4
7
4
0
3
9
0
.
7
7
6
4
3
9
0
.
3
5
3
6
6
6
0
.
8
3
4
4
4
6
0
.
3
4
1
3
2
5
0
.
4
8
1
0
7
7
1
.
2
1
6
4
5
9
1
.
1
4
0
7
4
1
1
.
3
8
9
4
8
3
-
0
.
0
4
1
5
3
C
l
u
st
e
r
1
-
0
.
2
0
6
1
-
0
.
3
3
7
5
8
-
0
.
1
5
3
7
7
-
0
.
3
6
2
8
-
0
.
1
4
8
4
-
0
.
2
0
9
1
6
-
0
.
5
2
8
9
-
0
.
4
9
5
9
7
-
0
.
6
0
4
1
2
0
.
0
1
8
0
5
7
T
ab
le
5
.
E
u
clid
ea
n
d
is
tan
ce
to
in
itial c
en
tr
o
id
C
r
i
t
e
r
i
a
C
e
n
t
r
o
i
d
0
C
e
n
t
r
o
i
d
1
C
r
i
t
1
2
.
0
8
3
3
0
4
5
.
7
9
3
3
C
r
i
t
2
3
.
1
1
2
8
2
3
5
.
6
6
6
3
1
7
C
r
i
t
3
2
.
8
1
9
3
3
3
5
.
6
2
3
0
0
8
C
r
i
t
4
.
.
2
.
0
9
1
7
5
3
5
.
6
8
6
9
4
8
C
r
i
t
3
3
6
.
5
2
0
2
0
7
3
.
5
5
2
6
4
T
ab
le
6
.
C
lu
s
ter
m
em
b
e
r
s
h
i
p
o
f
K
-
m
ea
n
s
C
l
u
st
e
r
To
t
a
l
mem
b
e
r
(
n
)
C
r
i
t
e
r
i
a
C0
10
0
,
1
,
2
,
3
,
4
,
5
,
6
,
7
,
8
,
9
C1
23
1
0
,
1
1
,
1
2
,
1
3
,
1
4
,
1
5
,
1
6
,
1
7
,
1
8
,
1
9
,
2
0
,
2
1
,
2
2
,
2
3
,
2
4
,
2
7
,
2
8
,
3
0
,
3
1
,
3
2
3
.
3
.
2
.
F
uzzy
C
-
m
ea
ns
T
h
e
clu
s
ter
in
g
p
r
o
ce
d
u
r
e
in
FC
M
g
en
er
ates
m
u
ltip
le
p
r
im
ar
y
o
u
tp
u
ts
th
at
d
ep
ict
th
e
g
en
er
ate
d
clu
s
ter
s
tr
u
ctu
r
e.
E
ac
h
d
ata
p
o
in
t'
s
m
em
b
er
s
h
ip
m
atr
ix
(
)
,
clu
s
ter
ce
n
tr
o
id
v
alu
es,
o
b
jectiv
e
f
u
n
ctio
n
v
alu
es,
an
d
clu
s
ter
lab
els ar
e
am
o
n
g
t
h
ese
o
u
tp
u
ts
[
6
0
]
–
[
6
5
]
.
i)
Me
m
b
er
s
h
ip
m
atr
ix
(
)
:
t
h
e
m
em
b
er
s
h
ip
m
atr
ix
(
)
f
o
r
3
3
d
ata
p
o
in
ts
in
th
e
FC
M
alg
o
r
ith
m
,
in
d
icatin
g
th
e
m
em
b
er
s
h
i
p
lev
els
to
war
d
two
clu
s
ter
s
,
i
s
d
is
p
lay
ed
in
T
ab
le
7
.
C
lu
s
ter
_
0
h
as
a
m
em
b
er
s
h
ip
o
f
0
.
7
7
8
3
0
3
f
o
r
d
ata
p
o
in
t
0
,
wh
ile
clu
s
ter
_
1
h
as
a
m
em
b
er
s
h
ip
o
f
0
.
2
2
1
6
9
7
,
s
u
g
g
esti
n
g
a
g
r
ea
ter
co
r
r
elatio
n
with
clu
s
ter
_
0
.
Fo
r
d
ata
p
o
in
t
3
2
,
clu
s
ter
_
0
is
0
.
3
2
3
1
0
5
,
an
d
clu
s
ter
_
1
is
0
.
6
7
6
8
9
5
,
s
u
g
g
esti
n
g
a
s
tr
o
n
g
er
co
r
r
elat
io
n
with
clu
s
ter
_
1
[
6
6
]
.
E
ac
h
co
lu
m
n
s
u
m
s
to
1
,
an
d
d
ef
u
zz
if
icatio
n
is
u
s
ed
to
d
eter
m
in
e
th
e
f
in
al
clu
s
ter
lab
el:
lab
el
0
if
clu
s
t
er
_
0
m
em
b
er
s
h
i
p
is
h
ig
h
er
,
an
d
lab
el
1
if
clu
s
ter
_
1
m
em
b
e
r
s
h
ip
is
h
ig
h
e
r
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J Ar
tif
I
n
tell
I
SS
N:
2252
-
8
9
3
8
C
o
mp
a
r
is
o
n
a
lg
o
r
ith
ms w
ith
o
p
timiz
a
tio
n
fo
r
clu
s
teri
n
g
mu
lti
-
crit
eria
(
I
d
a
Mu
lya
d
i
)
3483
T
ab
le
7
.
Me
m
b
er
s
h
ip
(
)
an
d
lab
el
C
l
u
st
e
r
0
1
2
3
4
5
32
C
l
u
st
e
r
_
0
0
.
7
7
8
3
0
3
0
.
7
1
4
7
8
5
0
.
7
3
5
5
8
5
0
.
7
7
6
1
4
9
0
.
6
9
8
8
7
3
0
.
7
3
7
6
3
1
0
.
3
2
3
1
0
5
C
l
u
st
e
r
_
1
0
.
2
2
1
6
9
7
0
.
2
8
5
2
1
5
0
.
2
6
4
4
1
5
0
.
2
2
3
8
5
1
0
.
3
0
1
1
2
7
0
.
2
6
2
3
6
9
0
.
6
7
6
8
9
5
La
b
e
l
0
0
0
0
0
0
1
ii)
C
lu
s
ter
ce
n
tr
o
id
s
:
t
h
e
weig
h
te
d
s
u
m
o
f
F1
v
alu
es,
b
ased
o
n
th
e
s
q
u
ar
e
o
f
t
h
e
m
e
m
b
er
s
h
i
p
d
e
g
r
ee
s
,
is
d
iv
id
ed
b
y
t
h
e
to
tal
s
q
u
a
r
e
o
f
th
e
m
em
b
er
s
h
i
p
s
.
T
o
d
eter
m
i
n
e
th
e
ce
n
tr
o
id
v
alu
e
o
f
clu
s
te
r
_
0
f
o
r
f
ea
tu
r
e
F1
in
FC
M
(
m
=
2
)
.
T
h
e
f
u
zz
y
weig
h
ted
av
e
r
ag
e
o
f
F1
in
clu
s
ter
_
0
is
r
ep
r
esen
ted
b
y
th
e
r
esu
ltin
g
ce
n
tr
o
id
v
al
u
e
0
.
2
7
5
2
7
2
.
0
.
1
=
(
(
0
.
90
3
0
)
2
×
0
.
4
3
7
96
)
+
(
(
0
.
0
97
0
)
2
×
0
.
2
1
8
98
)
+
⋯
+
(
(
0
.
90
3
)
2
×
0
.
4
3
7
96
)
(
0
.
90
3
0
)
2
+
(
0
.
90
3
)
2
+
⋯
+
(
0
.
0
97
)
2
=
0
.
275272
T
ab
le
8
p
r
esen
ts
th
e
ce
n
tr
o
id
v
alu
es
o
f
th
e
two
clu
s
ter
s
d
e
r
iv
ed
b
y
f
u
zz
y
C
-
m
ea
n
s
.
E
ac
h
n
u
m
b
er
f
r
o
m
F1
to
F1
6
d
e
n
o
tes
th
e
f
u
zz
y
weig
h
ted
av
er
a
g
e
o
f
ea
ch
f
ea
tu
r
e
in
a
class
if
icatio
n
.
C
lu
s
ter
_
0
ex
h
ib
its
ce
n
tr
o
id
v
al
u
es
th
at
ar
e
p
r
ed
o
m
in
an
tly
p
o
s
itiv
e,
in
clu
d
in
g
F1
=
0
.
2
7
5
2
7
2
an
d
F4
=
0
.
4
5
4
9
5
9
.
T
h
is
s
ig
n
if
ies
th
at
th
e
d
ata
in
s
id
e
t
h
is
clu
s
ter
ar
e
c
o
m
p
ar
ati
v
ely
ab
o
v
e
av
er
ag
e
,
ass
u
m
in
g
th
e
d
ata
h
av
e
b
ee
n
n
o
r
m
alize
d
.
C
o
n
v
er
s
ely
,
clu
s
ter
_
1
ex
h
ib
its
a
p
r
ed
o
m
in
an
t
n
eg
ativ
e
ce
n
tr
o
i
d
v
a
lu
e,
s
u
ch
as
F1
=
-
0
.
2
1
0
5
5
an
d
F4
=
-
0
.
3
5
0
1
7
.
T
h
is
s
ig
n
if
ies th
at
th
e
d
at
a
with
in
th
is
clu
s
ter
is
s
u
b
p
ar
.
T
ab
le
8
.
C
en
tr
o
id
clu
s
ter
C
l
u
st
e
r
F1
F2
F3
F4
F
5
…
F
1
6
C
l
u
st
e
r
_
0
0
.
2
7
5
2
7
2
0
.
4
3
7
3
5
5
0
.
1
9
9
1
4
5
0
.
4
5
4
9
5
9
0
.
1
4
7
7
5
4
0
.
0
1
6
0
6
1
C
l
u
st
e
r
_
1
-
0
.
2
1
0
5
5
-
0
.
3
2
5
1
1
-
0
.
1
5
8
0
5
-
0
.
3
5
0
1
7
-
0
.
0
9
5
9
3
-
0
.
0
7
2
7
9
iii)
T
h
e
o
b
jectiv
e
f
u
n
ctio
n
v
al
u
e
:
T
ab
le
9
s
h
o
ws
th
at
t
h
e
f
u
zz
y
C
-
m
ea
n
s
’
o
b
jectiv
e
f
u
n
ctio
n
v
alu
e
d
r
o
p
p
ed
f
r
o
m
3
0
9
.
6
7
2
2
2
2
in
th
e
f
ir
s
t
iter
atio
n
to
2
6
4
.
2
4
2
7
2
4
in
th
e
s
ec
o
n
d
,
d
em
o
n
s
tr
atin
g
n
o
tab
le
im
p
r
o
v
em
e
n
ts
in
th
e
m
em
b
er
s
h
ip
m
atr
ix
(
)
an
d
ce
n
tr
o
id
ad
j
u
s
tm
en
ts
.
T
h
e
v
alu
e
d
ec
lin
e
s
lo
wed
an
d
s
tab
ilized
af
ter
th
e
th
ir
d
iter
atio
n
.
I
ts
v
alu
e
r
ea
ch
ed
2
5
9
.
3
5
2
4
1
9
at
iter
atio
n
3
3
,
s
u
g
g
esti
n
g
th
at
th
e
o
p
tim
izatio
n
p
r
o
ce
s
s
h
as
s
tab
ilized
an
d
th
e
al
g
o
r
ith
m
is
alm
o
s
t c
o
n
v
er
g
en
t.
T
ab
le
9
.
I
ter
atio
n
an
d
o
b
jectiv
e
f
u
n
ctio
n
I
t
e
r
a
t
i
o
n
O
b
j
e
c
t
i
v
e
f
u
n
c
t
i
o
n
(
Jm)
1
3
0
9
.
6
7
2
2
2
2
2
2
6
4
.
2
4
2
7
2
4
3
2
6
3
.
3
4
8
5
7
4
4
2
6
2
.
8
2
7
0
3
3
5…
2
6
2
.
2
9
1
9
5
4
33
2
5
9
.
3
5
2
4
1
9
T
h
e
o
b
jectiv
e
f
u
n
ctio
n
(
)
is
d
e
ter
m
in
ed
b
y
s
u
m
m
in
g
th
e
E
u
c
lid
ea
n
d
is
tan
ce
(
)
f
o
r
ea
c
h
r
o
w
an
d
th
e
m
em
b
er
s
h
ip
m
atr
i
x
(
)
ac
co
r
d
in
g
to
(
5
)
.
T
h
is
p
r
o
ce
d
u
r
e
is
ex
ec
u
ted
f
o
r
all
d
ata
a
n
d
all
clu
s
ter
s
.
T
h
e
cu
m
u
lativ
e
f
in
d
in
g
s
y
ield
a
to
t
al
o
b
jectiv
e
f
u
n
ctio
n
v
alu
e
o
f
3
0
9
.
6
7
2
2
2
2
in
th
at
iter
atio
n
.
=
(
0
.
778303
)
2
(
0
.
43796
−
0
.
275272
)
2
+
(
0
.
221697
)
2
(
0
.
21898
−
(
−
0
.
21055
)
)
2
+
.
.
)
=
309
.
672222
iv
)
L
ab
el
clu
s
ter
:
T
ab
le
1
0
p
r
esen
ts
th
e
co
n
clu
s
iv
e
o
u
tco
m
es
o
f
d
ata
clu
s
ter
in
g
em
p
lo
y
in
g
FC
M
s
u
b
s
eq
u
en
t
to
th
e
d
e
f
u
zz
if
icatio
n
p
r
o
ce
d
u
r
e.
C
lu
s
ter
C
0
co
m
p
r
is
es
1
0
d
ata
p
o
in
ts
(
in
d
ices
0
–
9
)
,
r
e
p
r
esen
tin
g
th
e
d
ata
with
th
e
h
ig
h
est
af
f
iliatio
n
to
C
lu
s
ter
_
0
.
C
lu
s
ter
C
1
c
o
m
p
r
is
es
2
3
d
ata
p
o
in
ts
(
in
d
ices
1
0
–
2
4
,
2
7
,
2
8
,
3
0
,
3
1
,
an
d
3
2
)
,
ex
h
ib
itin
g
a
g
r
ea
ter
af
f
in
ity
f
o
r
C
lu
s
ter
_
1
.
T
ab
le
1
0
.
C
lu
s
ter
m
em
b
er
s
h
ip
o
f
f
u
z
zy
C
-
m
ea
n
s
C
l
u
st
e
r
To
t
a
l
mem
b
e
r
(
n
)
C
r
i
t
e
r
i
a
C0
11
0
,
1
,
2
,
3
,
4
,
5
,
6
,
7
,
8
,
9
,
1
5
C1
22
1
0
,
1
1
,
1
2
,
1
3
,
1
4
,
1
6
,
1
7
,
1
8
,
1
9
,
2
0
,
2
1
,
2
2
,
2
3
,
2
4
,
2
7
,
2
8
,
3
0
,
3
1
,
3
2
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8
9
3
8
I
n
t J Ar
tif
I
n
tell
,
Vo
l.
15
,
No
.
4
,
Au
g
u
s
t
20
26
:
3
4
7
6
-
3
4
9
1
3484
W
ith
1
1
d
ata
p
o
in
ts
in
clu
s
ter
C
0
an
d
2
2
in
clu
s
ter
C
1
,
T
a
b
le
1
1
d
em
o
n
s
tr
ates
th
at
b
o
th
K
-
m
ea
n
s
an
d
FC
M
g
en
er
ate
th
e
s
am
e
clu
s
ter
s
tr
u
ctu
r
e.
T
h
e
p
r
i
m
ar
y
d
is
tin
ctio
n
lies
in
th
e
ass
ig
n
m
en
t
m
ec
h
an
is
m
:
FC
M
u
s
e
s
s
o
f
t
as
s
ig
n
m
en
t,
wh
ich
allo
ws
d
ata
p
o
in
ts
to
h
av
e
m
em
b
er
s
h
i
p
v
alu
es
in
m
u
ltip
le
clu
s
ter
s
s
im
u
ltan
eo
u
s
ly
,
m
ak
i
n
g
it
m
o
r
e
ap
p
r
o
p
r
iate
f
o
r
d
ata
with
am
b
ig
u
o
u
s
clu
s
ter
b
o
u
n
d
ar
ie
s
,
wh
er
ea
s
K
-
m
ea
n
s
u
tili
ze
s
h
ar
d
ass
ig
n
m
en
t.
T
ab
le
1
1
.
A
n
aly
s
is
co
m
p
ar
is
o
n
h
y
b
r
id
elb
o
w+
s
ilh
o
u
ette
,
K
-
m
ea
n
s
,
an
d
FC
M
C
o
m
p
a
r
i
so
n
C
l
u
st
e
r
N
u
mb
e
r
o
f
m
e
m
b
e
r
s
C
r
i
t
e
r
i
a
H
y
b
r
i
d
e
l
b
o
w
+
si
l
h
o
u
e
t
t
e
+
K
-
m
e
a
n
s
C0
10
0
,
1
,
2
,
3
,
4
,
5
,
6
,
7
,
8
,
9
C1
23
1
0
,
1
1
,
1
2
,
1
3
,
1
4
,
1
5
,
1
6
,
1
7
,
1
8
,
1
9
,
2
0
,
2
1
,
2
2
,
2
3
,
2
4
,
2
7
,
2
8
,
3
0
,
3
1
,
3
2
H
y
b
r
i
d
e
l
b
o
w
+
si
l
h
o
u
e
t
t
e
+
f
u
z
z
y
C
-
mea
n
s
C0
11
0
,
1
,
2
,
3
,
4
,
5
,
6
,
7
,
8
,
9
,
1
5
C1
22
1
0
,
1
1
,
1
2
,
1
3
,
1
4
,
1
6
,
1
7
,
1
8
,
1
9
,
2
0
,
2
1
,
2
2
,
2
3
,
2
4
,
2
7
,
2
8
,
3
0
,
3
1
,
3
2
T
h
is
s
tu
d
y
clar
if
ies
th
e
f
ac
to
r
s
th
at
af
f
ec
t
s
tu
d
en
t
g
r
ad
u
ati
o
n
in
ter
m
s
o
f
eth
ics
an
d
ju
s
tice.
T
wo
clu
s
ter
s
ar
e
p
r
o
d
u
ce
d
b
y
th
is
s
tu
d
y
:
cl
u
s
ter
1
(
in
f
l
u
en
tial
n
o
n
-
ac
ad
em
ic
c
r
iter
ia)
a
n
d
clu
s
ter
0
(
v
er
y
i
n
f
lu
en
tial
ac
ad
em
ic
cr
iter
ia)
.
Dem
o
g
r
a
p
h
ic
an
d
s
o
cio
ec
o
n
o
m
ic
f
ac
to
r
s
ar
e
p
o
s
itio
n
ed
as
s
u
p
p
lem
en
tar
y
s
ettin
g
s
r
ath
er
th
an
as th
e
p
r
im
a
r
y
f
o
u
n
d
atio
n
f
o
r
clu
s
ter
in
g
o
r
d
eter
m
in
in
g
a
cr
iter
ia
.
3
.
4
.
Dept
h o
f
co
m
pa
riso
n us
ing
K
-
m
edo
ids
,
DB
SCAN,
s
pect
ra
l c
lus
t
er
,
G
M
M
,
a
nd
DL
T
a
b
le
1
2
s
h
o
ws
t
h
e
r
es
u
lts
o
f
t
h
e
c
o
m
p
a
r
is
o
n
a
n
al
y
s
is
o
f
K
-
m
ea
n
s
,
FC
M,
K
-
m
e
d
o
id
s
,
DB
SC
AN,
a
n
d
Sp
e
ct
r
al
wit
h
th
e
e
lb
o
w
+
s
il
h
o
u
et
te
h
y
b
r
id
.
K
-
m
ea
n
s
,
FC
M
,
GM
M,
a
n
d
DL
c
lu
s
te
r
i
n
g
m
et
h
o
d
s
y
i
el
d
t
h
e
s
a
m
e
r
es
u
lts
:
a
s
m
a
ll
cl
u
s
t
e
r
C
0
a
n
d
a
l
ar
g
e
cl
u
s
te
r
C
1
.
Fo
r
c
en
tr
o
id
-
b
ase
d
te
c
h
n
iq
u
es,
t
h
e
d
at
as
et
s
tr
u
ct
u
r
e
is
q
u
it
e
o
b
v
io
u
s
.
A
m
o
r
e
e
q
u
al
d
is
t
r
i
b
u
ti
o
n
o
f
K
-
m
ed
o
id
s
s
h
i
f
ts
s
o
m
e
d
a
ta
f
r
o
m
th
e
b
ig
cl
u
s
te
r
t
o
C
0
,
s
t
r
e
n
g
t
h
e
n
i
n
g
its
r
esis
ta
n
c
e
t
o
o
u
t
lie
r
s
.
Sp
ec
tr
al
cl
u
s
t
er
i
n
g
wit
h
i
n
v
e
r
te
d
clu
s
te
r
s
e
m
p
h
asiz
es
d
at
a
s
i
m
il
ar
i
ty
a
n
d
g
l
o
b
a
l
co
n
n
e
cte
d
n
ess
[
6
7
]
–
[
7
1
]
.
D
B
SC
AN
g
e
n
e
r
a
tes
a
l
ar
g
e
n
u
m
b
e
r
o
f
m
in
o
r
cl
u
s
t
er
s
t
h
at
ar
e
n
o
t
d
i
r
ec
tl
y
co
m
p
a
r
a
b
le
t
o
tw
o
m
a
jo
r
cl
u
s
t
er
s
b
u
t
a
r
e
u
s
e
f
u
l
f
o
r
d
e
tec
ti
n
g
o
u
tli
er
s
a
n
d
lo
ca
l
d
e
n
s
i
ty
s
t
r
u
c
tu
r
es
[
7
2
]
–
[
7
6
]
.
Usi
n
g
th
e
K
-
m
ea
n
s
,
FC
M,
K
-
m
ed
o
id
s
,
DB
SC
AN,
s
p
ec
tr
al
clu
s
ter
in
g
,
GM
M,
an
d
DL
a
lg
o
r
ith
m
s
b
ased
o
n
th
e
d
is
tan
ce
o
f
d
at
a
to
th
e
ce
n
tr
o
id
an
d
th
e
d
is
tan
ce
b
etwe
en
ce
n
tr
o
id
s
,
Fig
u
r
e
3
illu
s
tr
ates
th
e
d
if
f
er
en
ce
s
b
etwe
en
th
e
h
y
b
r
i
d
elb
o
w
an
d
s
ilh
o
u
ette.
K
-
m
e
an
s
,
f
u
zz
y
C
-
m
ea
n
s
,
GM
M,
an
d
DL
clu
s
ter
in
g
ar
e
th
e
m
o
s
t
s
tr
aig
h
tf
o
r
war
d
m
eth
o
d
s
.
T
h
e
ce
n
tr
o
i
d
s
ar
e
q
u
ite
clea
r
ly
s
ep
ar
ate
d
in
a
ll
f
o
u
r
tec
h
n
iq
u
es
.
No
n
eth
eless
,
GM
M
an
d
K
-
m
e
an
s
ap
p
ea
r
to
b
e
th
e
m
o
s
t
p
r
a
ctica
l
o
p
tio
n
s
if
we
m
u
s
t
s
elec
t
th
e
m
o
s
t
v
is
u
ally
s
tab
le,
as
th
ey
y
iel
d
two
c
o
m
p
ar
ativ
ely
d
is
tin
ct
an
d
f
air
l
y
ex
p
licab
le
g
r
o
u
p
s
.
Fo
llo
win
g
a
c
o
m
p
ar
is
o
n
o
f
ea
ch
alg
o
r
ith
m
'
s
clu
s
ter
in
g
p
er
f
o
r
m
an
ce
in
T
ab
le
1
2
,
a
cr
iter
io
n
s
elec
tio
n
tech
n
iq
u
e
was
a
p
p
lied
to
i
d
en
tify
th
e
m
o
s
t
in
f
lu
en
tial
f
ac
to
r
s
.
T
h
e
Sh
ap
ley
ad
d
itiv
e
ex
p
lan
atio
n
s
(
SHAP)
m
eth
o
d
was
u
s
ed
to
ev
alu
ate
th
e
s
ig
n
if
ican
ce
o
f
th
ese
f
ac
t
o
r
s
in
s
h
ap
in
g
clu
s
ter
in
g
o
u
tco
m
es.
T
ab
le
1
2
.
C
o
m
p
ar
is
o
n
a
n
aly
s
is
o
f
K
-
m
ea
n
s
,
FC
M,
K
-
m
ed
o
i
d
s
,
DB
SC
AN,
s
p
ec
tr
al,
GM
M
,
an
d
DL
with
elb
o
w
+
s
ilh
o
u
ette
C
o
m
p
a
r
i
so
n
C
l
u
st
e
r
N
u
mb
e
r
o
f
memb
e
r
s
C
r
i
t
e
r
i
a
H
y
b
r
i
d
e
l
b
o
w
+
si
l
h
o
u
e
t
t
e
+
K
-
m
e
a
n
s
C0
10
0
,
1
,
2
,
3
,
4
,
5
,
6
,
7
,
8
,
9
C1
23
1
0
,
1
1
,
1
2
,
1
3
,
1
4
,
1
5
,
1
6
,
1
7
,
1
8
,
1
9
,
2
0
,
2
1
,
2
2
,
2
3
,
2
4
,
2
5
,
2
6
,
2
7
,
2
8
,
3
0
,
3
1
,
3
2
H
y
b
r
i
d
e
l
b
o
w
+
si
l
h
o
u
e
t
t
e
+
F
u
z
z
y
C
-
m
e
a
n
s
C0
11
0
,
1
,
2
,
3
,
4
,
5
,
6
,
7
,
8
,
9
,
1
5
C1
22
1
0
,
1
1
,
1
2
,
1
3
,
1
4
,
1
6
,
1
7
,
1
8
,
1
9
,
2
0
,
2
1
,
2
2
,
2
3
,
2
4
,
2
5
,
2
6
,
2
7
,
2
8
,
3
0
,
3
1
,
3
2
H
y
b
r
i
d
e
l
b
o
w
+
si
l
h
o
u
e
t
t
e
+
K
-
m
e
d
o
i
d
s
C0
22
0
,
1
,
2
,
3
,
5
,
8
,
1
0
,
1
1
,
1
3
,
1
5
,
1
6
,
1
9
,
2
0
,
2
2
,
2
3
,
2
4
,
2
5
,
2
6
,
2
8
,
2
9
,
3
1
,
3
2
C1
11
4
,
6
,
7
,
9
,
1
2
,
1
4
,
1
7
,
1
8
,
2
1
,
2
7
,
3
0
H
y
b
r
i
d
e
l
b
o
w
+
si
l
h
o
u
e
t
t
e
+
s
p
e
c
t
r
a
l
c
l
u
s
t
e
r
i
n
g
C0
32
1
0
,
1
1
,
1
2
,
1
3
,
1
4
,
1
5
,
1
6
,
1
7
,
1
8
,
1
9
,
2
0
,
2
1
,
2
2
,
2
3
,
2
4
,
2
5
,
2
6
,
2
7
,
2
8
,
2
9
,
3
0
,
3
1
,
3
2
C1
10
0
,
1
,
2
,
3
,
4
,
5
,
6
,
7
,
8
,
9
H
y
b
r
i
d
e
l
b
o
w
+
si
l
h
o
u
e
t
t
e
+
D
B
S
C
A
N
3
2
c
l
u
s
t
e
r
s
a
r
e
f
o
r
me
d
,
w
i
t
h
c
l
u
st
e
r
6
h
a
v
i
n
g
t
w
o
c
r
i
t
e
r
i
a
—
6
a
n
d
7
.
H
y
b
r
i
d
e
l
b
o
w
+
si
l
h
o
u
e
t
t
e
+
G
M
M
C0
10
0
,
1
,
2
,
3
,
4
,
5
,
6
,
7
,
8
,
9
C1
23
1
0
,
1
1
,
1
2
,
1
3
,
1
4
,
1
5
,
1
6
,
1
7
,
1
8
,
1
9
,
2
0
,
2
1
,
2
2
,
2
3
,
2
4
,
2
5
,
2
6
,
2
7
,
2
8
,
3
0
,
3
1
,
3
2
H
y
b
r
i
d
e
l
b
o
w
+
si
l
h
o
u
e
t
t
e
+
DL
C0
10
0
,
1
,
2
,
3
,
4
,
5
,
6
,
7
,
8
,
9
C1
23
1
0
,
1
1
,
1
2
,
1
3
,
1
4
,
1
5
,
1
6
,
1
7
,
1
8
,
1
9
,
2
0
,
2
1
,
2
2
,
2
3
,
2
4
,
2
5
,
2
6
,
2
7
,
2
8
,
3
0
,
3
1
,
3
2
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J Ar
tif
I
n
tell
I
SS
N:
2252
-
8
9
3
8
C
o
mp
a
r
is
o
n
a
lg
o
r
ith
ms w
ith
o
p
timiz
a
tio
n
fo
r
clu
s
teri
n
g
mu
lti
-
crit
eria
(
I
d
a
Mu
lya
d
i
)
3485
Fig
u
r
e
3
.
C
o
m
p
a
r
is
o
n
o
f
h
y
b
r
i
d
elb
o
w+
s
ilh
o
u
ette
u
s
in
g
K
-
m
ea
n
s
,
FC
M,
K
-
m
ed
o
id
s
,
s
p
ec
tr
al
clu
s
ter
in
g
,
DB
S
C
AN,
GM
M,
an
d
DL
clu
s
ter
in
g
T
h
e
r
a
n
g
e
o
f
in
f
lu
e
n
ce
f
r
o
m
s
tr
o
n
g
to
lo
w
is
illu
s
tr
a
ted
in
Fig
u
r
e
4
.
K
-
m
ea
n
s
is
q
u
ick
,
m
em
o
r
y
-
ef
f
icien
t,
a
n
d
s
till
g
e
n
er
ates
two
clu
s
ter
s
,
T
ab
le
1
3
d
em
o
n
s
tr
ates
th
at
it
is
th
e
m
o
s
t
b
alan
ce
d
m
eth
o
d
.
Alth
o
u
g
h
FC
Me
an
is
f
aster
th
an
K
-
m
ea
n
s
,
it
u
s
es
a
litt
le
m
o
r
e
m
em
o
r
y
.
B
ec
au
s
e
DB
SC
A
N
ca
n
o
n
ly
cr
ea
te
a
s
in
g
le
clu
s
ter
with
th
e
g
iv
e
n
p
ar
am
eter
s
,
it
is
n
o
t
a
d
v
is
ed
.
W
h
ile
K
-
m
ed
o
id
s
is
s
till
u
s
ef
u
l
b
u
t
n
ee
d
s
f
u
r
th
er
clu
s
ter
q
u
ality
ass
ess
m
en
t,
s
p
ec
tr
al
clu
s
ter
in
g
is
less
ef
f
ec
tiv
e
f
o
r
lar
g
e
d
ata.
Fig
u
r
e
4
.
Featu
r
e
im
p
o
r
ta
n
ce
r
an
k
in
g
SHAP
Evaluation Warning : The document was created with Spire.PDF for Python.