I
nte
rna
t
io
na
l J
o
urna
l o
f
E
lect
rica
l a
nd
Co
m
pu
t
er
E
ng
ineering
(
I
J
E
CE
)
Vo
l.
1
6
,
No
.
1
,
Feb
r
u
ar
y
20
2
6
,
p
p
.
297
~
310
I
SS
N:
2088
-
8
7
0
8
,
DOI
: 1
0
.
1
1
5
9
1
/ijece.
v
1
6
i
1
.
pp
2
9
7
-
3
1
0
297
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ij
ec
e.
ia
esco
r
e.
co
m
Studen
ts per
form
a
nce cluste
ring
f
o
r f
ut
ure
perso
na
lized in
lea
rning
virtua
l r
ea
lity
G
ha
lia
M
da
g
hri A
la
o
ui,
Ab
delha
m
id Z
o
uh
a
ir,
I
l
ha
m
e
K
ha
bb
a
chi
D
S
A
I
2
S
R
e
s
e
a
r
c
h
Te
a
m,
C
3
S
La
b
o
r
a
t
o
r
y
,
F
a
c
u
l
t
y
o
f
S
c
i
e
n
c
e
s
a
n
d
Te
c
h
n
o
l
o
g
i
e
s
o
f
Ta
n
g
i
e
r
,
A
b
d
e
l
ma
l
e
k
Essaâ
d
i
U
n
i
v
e
r
s
i
t
y
,
Te
t
o
u
a
n
,
M
o
r
o
c
c
o
Art
icle
I
nfo
AB
S
T
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
Ma
r
2
2
,
2
0
2
5
R
ev
is
ed
Oct
6
,
2
0
2
5
Acc
ep
ted
No
v
2
3
,
2
0
2
5
Th
is
st
u
d
y
i
n
v
e
stig
a
tes
fi
v
e
c
l
u
ste
rin
g
a
lg
o
rit
h
m
s
—
K
-
M
e
a
n
s,
G
a
u
ss
ian
m
ix
tu
re
m
o
d
e
l
(G
M
M
)
,
h
iera
rc
h
ica
l
c
lu
ste
rin
g
(HC)
,
k
-
m
e
d
o
id
s,
a
n
d
sp
e
c
tral
c
lu
ste
rin
g
—
a
p
p
li
e
d
to
st
u
d
e
n
t
p
e
rfo
rm
a
n
c
e
i
n
m
a
th
e
m
a
ti
c
s,
re
a
d
in
g
,
a
n
d
writi
n
g
to
su
p
p
o
r
t
t
h
e
d
e
v
e
lo
p
m
e
n
t
o
f
v
irt
u
a
l
re
a
li
ty
(
VR)
-
b
a
se
d
a
d
a
p
ti
v
e
lea
rn
i
n
g
sy
ste
m
s.
Clu
ste
r
q
u
a
li
ty
wa
s
a
ss
e
ss
e
d
u
sin
g
Da
v
ies
-
Bo
u
l
d
in
a
n
d
Ca
li
n
s
k
i
-
Ha
ra
b
a
sz
in
d
ice
s.
S
p
e
c
tral
c
lu
ste
rin
g
a
c
h
ie
v
e
d
th
e
b
e
s
t
re
su
lt
s
(DBI =
0
.
7
5
,
CHI =
1
3
2
2
),
f
o
ll
o
we
d
b
y
K
-
M
e
a
n
s
(DBI =
0
.
7
9
,
CHI =
1
3
9
8
),
wh
i
le
HC
d
e
m
o
n
str
a
ted
su
p
e
rio
r
ro
b
u
st
n
e
ss
to
o
u
tl
ie
rs.
Th
re
e
d
isti
n
c
t
stu
d
e
n
t
p
r
o
fil
e
s
—
b
e
g
in
n
e
r,
in
term
e
d
iate
,
a
n
d
a
d
v
a
n
c
e
d
—
e
m
e
rg
e
d
,
e
n
a
b
li
n
g
targ
e
te
d
a
d
a
p
t
iv
e
i
n
ter
v
e
n
ti
o
n
s.
S
u
p
e
rv
ise
d
c
las
sifiers
train
e
d
o
n
th
e
se
c
lu
ste
rs
re
a
c
h
e
d
u
p
to
9
9
%
a
c
c
u
ra
c
y
(lo
g
isti
c
re
g
re
ss
io
n
)
a
n
d
9
7
.
5
%
(
su
p
p
o
rt
v
e
c
t
o
r
m
a
c
h
in
e
(S
VM)
),
v
a
li
d
a
ti
n
g
t
h
e
d
isc
o
v
e
re
d
g
ro
u
p
i
n
g
s.
T
h
is
wo
rk
i
n
tro
d
u
c
e
s
a
n
o
v
e
l
,
d
a
ta
-
d
r
i
v
e
n
m
e
th
o
d
o
lo
g
y
in
teg
ra
ti
n
g
u
n
s
u
p
e
rv
ise
d
c
lu
ste
rin
g
with
su
p
e
rv
ise
d
p
re
d
i
c
ti
o
n
,
p
r
o
v
i
d
i
n
g
a
p
ra
c
ti
c
a
l
fra
m
e
wo
rk
f
o
r
d
e
sig
n
i
n
g
imm
e
rsiv
e
VR l
e
a
rn
in
g
e
n
v
ir
o
n
m
e
n
ts.
K
ey
w
o
r
d
s
:
Ad
ap
tiv
e
lear
n
in
g
s
y
s
tem
s
C
lu
s
ter
in
g
alg
o
r
ith
m
s
Sp
ec
tr
al
clu
s
ter
in
g
Stu
d
en
ts
p
er
f
o
r
m
an
ce
Vir
tu
al
r
ea
lity
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
:
Gh
alia
Md
ag
h
r
i
-
Alao
u
i
DSAI
2
S
R
esear
ch
T
ea
m
,
C
o
m
p
u
ter
Scien
ce
an
d
Sm
ar
t
Sy
s
te
m
s
L
ab
o
r
ato
r
y
,
FS
T
of
T
an
g
ie
r
,
Ab
d
elm
alek
E
s
s
aâ
d
i U
n
iv
er
s
ity
T
eto
u
an
,
M
o
r
o
c
c
o
E
m
ail:
g
h
alia.
m
d
ag
h
r
ialao
u
i
@
etu
.
u
ae
.
ac
.
m
a
1.
I
NT
RO
D
UCT
I
O
N
T
h
is
wo
r
k
aim
s
to
an
aly
ze
s
tu
d
en
t
p
er
f
o
r
m
an
ce
in
h
ig
h
er
ed
u
ca
tio
n
in
s
titu
tio
n
s
(
H
E
I
s
)
u
s
in
g
clu
s
ter
in
g
an
d
class
if
icatio
n
m
eth
o
d
s
to
in
f
o
r
m
v
ir
tu
al
r
e
ality
(
VR
)
-
b
ased
p
e
r
s
o
n
alize
d
lea
r
n
in
g
.
Per
s
o
n
alize
d
lear
n
in
g
r
ep
r
esen
ts
a
s
h
if
t
f
r
o
m
tr
ad
itio
n
al
o
n
e
-
s
ize
-
f
its
-
all
ap
p
r
o
ac
h
es
to
war
d
tailo
r
ed
i
n
s
tr
u
ctio
n
th
at
a
d
ap
ts
to
ea
ch
s
tu
d
en
t’
s
ab
ilit
ies,
p
r
ef
er
en
ce
s
,
an
d
n
e
ed
s
.
Vir
tu
al
r
ea
lity
(
VR
)
en
a
b
les
im
m
er
s
iv
e
lear
n
i
n
g
en
v
ir
o
n
m
en
ts
th
at
ca
n
ad
j
u
s
t
d
y
n
am
ically
to
lear
n
er
s
,
wh
il
e
clu
s
ter
in
g
alg
o
r
ith
m
s
allo
w
th
e
id
en
tific
atio
n
o
f
m
ea
n
in
g
f
u
l
s
tu
d
en
t
g
r
o
u
p
s
,
s
u
p
p
o
r
tin
g
a
d
ap
tiv
e
i
n
ter
v
en
ti
o
n
s
an
d
m
o
r
e
ef
f
icien
t,
cu
s
to
m
ized
ed
u
ca
tio
n
al
ex
p
er
ien
ce
s
.
Sev
er
al
s
tu
d
ies
h
a
v
e
a
p
p
lied
clu
s
ter
in
g
to
ed
u
ca
tio
n
al
p
er
s
o
n
aliza
tio
n
.
O
u
ass
if
et
a
l
.
[
1
]
u
s
ed
K
-
Me
an
s
o
n
e
n
g
ag
e
m
en
t
b
e
h
a
v
io
r
s
,
b
u
t
th
eir
a
p
p
r
o
ac
h
was
l
im
ited
to
a
s
in
g
le
alg
o
r
it
h
m
a
n
d
d
ataset,
with
n
o
co
n
n
ec
tio
n
t
o
Vah
d
at
et
a
l
.
[
2
]
p
r
o
v
id
ed
a
g
en
er
al
r
ev
iew
with
o
u
t
ex
p
er
im
e
n
tatio
n
.
Ša
r
ić
-
Gr
g
ić
et
a
l
.
[
3
]
an
aly
ze
d
b
e
h
av
io
r
s
in
an
in
tel
lig
en
t
tu
to
r
in
g
s
y
s
tem
,
b
u
t
r
es
u
lts
wer
e
co
n
f
in
ed
to
o
n
lin
e
l
ea
r
n
in
g
.
H
o
o
s
h
y
ar
et
a
l
.
[
4
]
in
tr
o
d
u
ce
d
th
e
PP
P
alg
o
r
ith
m
b
ased
o
n
p
r
o
cr
asti
n
atio
n
,
ef
f
ec
tiv
e
b
u
t
lim
ited
to
a
s
in
g
le
v
ar
iab
le.
Nav
ar
r
o
an
d
Ger
[
5
]
co
m
p
ar
e
d
alg
o
r
ith
m
s
o
n
lar
g
e
d
atasets
with
o
u
t
co
n
s
id
er
in
g
im
m
er
s
i
v
e
p
er
s
o
n
aliza
tio
n
.
DeFr
eitas
an
d
B
er
n
ar
d
[
6
]
e
v
alu
ated
K
-
Me
an
s
,
d
en
s
ity
-
b
as
ed
s
p
atial
clu
s
ter
in
g
o
f
ap
p
licatio
n
s
with
n
o
is
e
(
DB
SC
AN
)
,
an
d
b
alan
ce
d
iter
ativ
e
r
ed
u
cin
g
a
n
d
clu
s
ter
in
g
u
s
in
g
h
ier
ar
ch
ies (
B
I
R
C
H)
,
co
n
f
ir
m
in
g
K
-
Me
a
n
s
’
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8
7
0
8
I
n
t J E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
1
6
,
No
.
1
,
Feb
r
u
ar
y
20
2
6
:
2
9
7
-
310
298
ef
f
ec
tiv
en
ess
b
u
t
o
n
ly
o
n
in
te
r
n
al
m
etr
ics.
Kr
ižan
ić
[
7
]
co
m
b
in
ed
clu
s
ter
in
g
an
d
d
ec
is
io
n
tr
ee
s
o
n
e
-
lear
n
in
g
lo
g
s
,
wh
ile
Vital
et
a
l
.
[
8
]
i
n
teg
r
ated
s
tatis
tical
an
aly
s
is
an
d
clu
s
ter
in
g
b
ased
o
n
s
o
cio
-
p
e
r
s
o
n
al
f
ac
to
r
s
.
R
ec
en
t
ad
v
an
ce
s
in
p
r
ed
ictiv
e
m
o
d
eli
n
g
f
u
r
t
h
er
h
ig
h
lig
h
t
th
e
p
o
te
n
tial
o
f
s
o
p
h
is
ticated
ap
p
r
o
ac
h
es
f
o
r
c
o
m
p
lex
n
o
n
lin
ea
r
p
atter
n
s
.
J
in
et
a
l
.
[
9
]
a
p
p
lied
n
eu
r
a
l
n
etwo
r
k
s
to
ca
p
tu
r
e
tem
p
o
r
al
d
ep
en
d
en
cies
in
tr
ad
in
g
v
o
lu
m
es,
an
d
J
in
et
a
l
.
[
1
0
]
u
s
ed
th
e
s
am
e
ap
p
r
o
ac
h
to
f
o
r
ec
ast
co
m
m
o
d
ity
p
r
ic
es.
J
in
an
d
Xu
[
1
1
]
d
em
o
n
s
tr
ated
th
e
ef
f
ec
tiv
e
n
e
s
s
o
f
Gau
s
s
ian
p
r
o
ce
s
s
r
eg
r
ess
io
n
(
GPR
)
with
B
ay
es
ian
o
p
tim
izatio
n
f
o
r
p
r
ed
ictin
g
s
ilv
er
p
r
ices,
an
d
J
in
an
d
Xu
[
1
2
]
ap
p
lied
th
is
m
eth
o
d
t
o
th
er
m
al
co
al
p
r
ice
s
.
J
in
an
d
Xu
[
1
3
]
em
p
lo
y
ed
g
r
a
p
h
ical
m
o
d
els,
in
clu
d
in
g
d
ir
ec
ted
ac
y
clic
g
r
ap
h
s
(
DAGs)
,
to
u
n
co
v
er
ca
u
s
al
s
tr
u
ctu
r
es
in
m
u
ltiv
ar
iate
ec
o
n
o
m
ic
d
ata,
w
h
ile
Xu
[
1
4
]
ex
ten
d
ed
th
is
ty
p
e
o
f
an
aly
s
is
.
Xu
[
1
5
]
s
h
o
wed
th
at
en
s
em
b
le
an
d
co
m
p
o
s
ite
m
eth
o
d
s
im
p
r
o
v
e
p
r
ed
ictio
n
r
o
b
u
s
tn
ess
f
o
r
a
g
r
icu
ltu
r
al
co
m
m
o
d
ities
,
an
d
X
u
an
d
Z
h
an
g
[
1
6
]
co
n
f
ir
m
e
d
th
ese
b
en
e
f
its
f
o
r
f
in
an
cial
in
d
ices.
I
n
s
p
ir
ed
b
y
th
ese
wo
r
k
s
,
o
u
r
s
tu
d
y
le
v
er
ag
es
clu
s
ter
in
g
an
d
class
if
icatio
n
tech
n
iq
u
es to
an
aly
ze
s
tu
d
en
t p
er
f
o
r
m
a
n
ce
an
d
g
u
id
e
a
d
ap
tiv
e
VR
-
b
ased
lear
n
in
g
.
Desp
ite
th
ese
ad
v
an
ce
s
,
p
r
ev
io
u
s
r
esear
ch
m
ain
ly
ap
p
lied
s
in
g
le
clu
s
ter
in
g
m
eth
o
d
s
to
e
d
u
ca
tio
n
al
d
ata,
with
lim
ited
v
alid
atio
n
an
d
litt
le
co
n
n
ec
tio
n
to
im
m
e
r
s
iv
e
p
er
s
o
n
aliza
tio
n
.
T
h
is
s
tu
d
y
ad
d
r
ess
es
th
es
e
g
ap
s
b
y
ev
al
u
atin
g
f
iv
e
clu
s
ter
in
g
alg
o
r
ith
m
s
(
K
-
Me
an
s
,
Gau
s
s
ian
m
ix
tu
r
e
m
o
d
el
(
GM
M)
,
h
ier
ar
ch
ical,
K
-
m
ed
o
id
s
,
an
d
s
p
ec
tr
al
clu
s
ter
in
g
)
,
q
u
an
titativ
ely
v
alid
at
in
g
th
e
d
is
co
v
er
e
d
clu
s
ter
s
th
r
o
u
g
h
s
u
p
er
v
is
ed
class
if
ier
s
,
an
d
id
en
tify
in
g
in
ter
p
r
etab
le
s
tu
d
en
t
p
r
o
f
iles
—
b
eg
in
n
er
,
in
ter
m
ed
iate,
an
d
ad
v
an
ce
d
—
to
in
f
o
r
m
ad
ap
tiv
e
in
ter
v
e
n
tio
n
s
.
T
h
e
co
n
tr
ib
u
tio
n
s
o
f
th
is
s
tu
d
y
ar
e
s
u
m
m
ar
ized
in
a
s
in
g
le
p
ar
ag
r
ap
h
as
f
o
llo
ws:
f
ir
s
t,
a
s
y
s
tem
atic
co
m
p
a
r
is
o
n
o
f
f
i
v
e
clu
s
ter
in
g
alg
o
r
ith
m
s
o
n
H
E
I
s
tu
d
en
t
p
er
f
o
r
m
a
n
ce
d
ata;
s
ec
o
n
d
,
v
alid
atio
n
o
f
clu
s
ter
s
u
s
in
g
s
u
p
er
v
is
ed
clas
s
if
icatio
n
m
o
d
els
to
en
s
u
r
e
q
u
an
titativ
e
r
o
b
u
s
tn
ess
;
an
d
th
ir
d
,
id
en
tific
atio
n
o
f
in
ter
p
r
etab
le
s
tu
d
en
t
p
r
o
f
iles
to
g
u
id
e
ad
ap
tiv
e
VR
-
b
ased
lear
n
in
g
an
d
p
r
o
v
i
d
e
a
p
r
ac
tical
f
r
am
ewo
r
k
f
o
r
im
m
er
s
iv
e,
p
er
s
o
n
alize
d
ed
u
c
atio
n
.
T
h
e
r
em
ain
d
e
r
o
f
th
e
p
a
p
er
is
o
r
g
an
ized
as
f
o
llo
ws:
s
ec
tio
n
2
r
ev
iews
r
elate
d
r
esear
ch
o
n
clu
s
ter
in
g
m
eth
o
d
s
in
e
d
u
ca
tio
n
al
d
ata
m
in
in
g
.
Sectio
n
3
p
r
esen
ts
o
u
r
m
eth
o
d
o
lo
g
y
,
in
clu
d
in
g
p
r
ep
r
o
ce
s
s
in
g
,
d
ataset
attr
ib
u
tes,
s
y
s
tem
ar
ch
itectu
r
e,
clu
s
ter
in
g
alg
o
r
ith
m
s
,
an
d
e
v
alu
atio
n
m
etr
ics.
Sectio
n
4
r
ep
o
r
ts
ex
p
er
im
en
tal
r
esu
lts
,
co
m
p
ar
es
class
if
icatio
n
m
o
d
els,
an
aly
ze
s
s
tu
d
e
n
t
g
r
o
u
p
s
,
an
d
ev
alu
ates
clu
s
t
er
in
g
p
e
r
f
o
r
m
an
ce
.
Fin
ally
,
s
ec
tio
n
5
co
n
clu
d
es
an
d
o
u
tlin
es
f
u
t
u
r
e
d
ir
ec
tio
n
s
,
in
clu
d
i
n
g
t
h
e
a
p
p
licatio
n
o
f
d
ee
p
lear
n
in
g
a
n
d
r
ea
l
-
tim
e
f
ee
d
b
ac
k
s
y
s
tem
s
.
2.
L
I
T
E
R
AT
U
RE
R
E
VI
E
W
2
.
1
.
Clus
t
er
ing
t
ec
hn
i
qu
e
in e
du
c
a
t
io
na
l da
t
a
m
ini
ng
C
lu
s
ter
in
g
tech
n
iq
u
es
h
av
e
b
ec
o
m
e
a
co
r
n
er
s
to
n
e
o
f
ed
u
ca
tio
n
al
d
ata
m
in
in
g
(
E
DM
)
,
e
n
ab
lin
g
th
e
id
en
tific
atio
n
o
f
m
ea
n
in
g
f
u
l
p
atter
n
s
in
s
tu
d
en
t
p
e
r
f
o
r
m
an
ce
,
en
g
ag
em
en
t,
an
d
b
e
h
a
v
io
r
.
E
a
r
ly
m
o
d
els
(
o
v
er
lay
,
f
u
zz
y
lo
g
ic,
B
ay
esian
n
etwo
r
k
s
)
p
r
o
v
id
e
d
s
o
lid
f
o
u
n
d
atio
n
s
b
u
t
r
em
ain
f
r
ag
m
en
ted
an
d
p
o
o
r
l
y
s
u
ited
to
ad
ap
tiv
e
lear
n
in
g
s
y
s
tem
s
[
1
7
]
.
R
ec
en
t
s
tu
d
ies
h
av
e
ap
p
lied
clu
s
ter
in
g
to
o
n
lin
e
lear
n
in
g
en
v
ir
o
n
m
en
ts
.
Šar
ić
-
Gr
g
ić
et
a
l.
[
3
]
p
er
f
o
r
m
e
d
clu
s
ter
in
g
o
f
s
tu
d
en
ts
b
ased
o
n
eig
h
t o
n
lin
e
b
eh
av
io
r
v
a
r
iab
les
in
an
in
tellig
en
t
tu
to
r
in
g
s
y
s
tem
(
AC
-
war
e
T
u
to
r
)
,
in
cl
u
d
in
g
p
r
ep
r
o
ce
s
s
in
g
,
d
im
en
s
io
n
ality
r
ed
u
ctio
n
,
clu
s
ter
in
g
,
an
d
p
o
s
t
-
test
p
er
f
o
r
m
an
ce
a
n
aly
s
is
,
an
d
cr
ea
t
ed
a
d
ec
is
io
n
tr
ee
f
o
r
h
u
m
a
n
in
ter
p
r
etatio
n
o
f
clu
s
ter
s
.
Ho
wev
er
,
th
e
ap
p
licatio
n
was
r
estricte
d
to
a
s
p
ec
if
ic
o
n
lin
e
s
y
s
tem
an
d
m
ay
n
o
t
g
en
er
alize
to
i
n
-
p
er
s
o
n
o
r
VR
lear
n
in
g
e
n
v
ir
o
n
m
en
ts
.
Ho
o
s
h
y
a
r
et
a
l.
[
4
]
d
ev
elo
p
ed
t
h
e
PP
P
alg
o
r
ith
m
to
p
r
ed
ict
s
tu
d
en
t
p
er
f
o
r
m
an
ce
ac
co
r
d
in
g
to
p
r
o
cr
asti
n
atio
n
b
eh
av
io
r
,
class
if
y
in
g
s
tu
d
en
ts
as
p
r
o
c
r
asti
n
ato
r
s
,
ca
n
d
i
d
ates,
o
r
non
-
p
r
o
cr
asti
n
ato
r
s
,
ac
h
iev
in
g
9
6
%
ac
cu
r
ac
y
with
m
u
ltip
le
class
if
ier
s
;
h
o
wev
er
,
t
h
is
ap
p
r
o
ac
h
f
o
cu
s
ed
m
ain
ly
o
n
p
r
o
c
r
asti
n
atio
n
,
li
m
itin
g
o
v
er
all
p
er
f
o
r
m
an
ce
p
r
ed
ictio
n
a
n
d
n
o
t
co
n
s
id
er
in
g
o
th
er
b
e
h
av
io
r
al
o
r
ac
ad
em
ic
v
ar
iab
les.
Nav
a
r
r
o
an
d
Ger
[
5
]
c
o
m
p
ar
e
d
d
if
f
er
en
t
clu
s
ter
in
g
alg
o
r
ith
m
s
o
n
a
lar
g
e
ed
u
ca
tio
n
al
d
ataset,
s
h
o
win
g
th
at
K
-
Me
an
s
an
d
p
ar
titi
o
n
in
g
ar
o
u
n
d
m
e
d
o
id
s
(
PAM
)
p
er
f
o
r
m
e
d
b
est
f
o
r
p
a
r
titi
o
n
in
g
.
At
th
e
s
am
e
tim
e,
d
iv
is
iv
e
an
aly
s
is
(
DI
ANA)
ex
ce
lled
in
h
ier
ar
ch
ical
clu
s
ter
in
g
,
th
o
u
g
h
th
e
s
tu
d
y
f
o
cu
s
ed
o
n
lar
g
e
d
atasets
with
o
u
t
ad
d
r
ess
in
g
VR
o
r
im
m
er
s
iv
e
p
er
s
o
n
al
izatio
n
an
d
d
id
n
o
t
tr
ac
k
in
d
iv
id
u
al
p
er
f
o
r
m
a
n
ce
.
Fu
s
ein
i
an
d
Miss
ah
[
1
8
]
co
n
f
ir
m
ed
th
e
d
o
m
in
a
n
ce
o
f
cl
u
s
ter
in
g
in
h
ig
h
e
r
ed
u
ca
tio
n
,
wh
ile
L
i
et
a
l
.
[
1
9
]
ap
p
lied
e
n
s
em
b
le
clu
s
ter
in
g
t
o
d
etec
t
t
y
p
ical
an
d
a
n
o
m
alo
u
s
b
eh
av
i
o
r
s
,
y
et
r
estricte
d
to
a
s
in
g
le
i
n
s
titu
tio
n
.
DeFr
eitas
an
d
B
er
n
ar
d
[
6
]
also
an
aly
ze
d
cl
u
s
ter
in
g
alg
o
r
ith
m
s
o
n
lear
n
in
g
m
an
a
g
em
en
t
s
y
s
tem
(
L
MS)
d
ata,
co
m
p
ar
in
g
K
-
Me
a
n
s
,
DB
SC
AN,
an
d
B
I
R
C
H,
with
K
-
Me
an
s
ac
h
iev
in
g
th
e
h
ig
h
est
Sil
h
o
u
ette
c
o
ef
f
icien
ts
;
lim
itatio
n
s
in
clu
d
ed
a
lack
o
f
ap
p
licatio
n
to
im
m
er
s
iv
e
s
y
s
tem
s
,
f
u
tu
r
e
p
er
f
o
r
m
a
n
c
e
p
r
ed
ictio
n
,
an
d
p
ed
ag
o
g
ical
in
ter
p
r
etatio
n
.
Kr
ižan
ić
[
7
]
ap
p
lied
d
ata
m
in
in
g
to
e
-
lear
n
in
g
lo
g
s
f
r
o
m
a
C
r
o
atian
u
n
iv
er
s
ity
,
u
s
in
g
clu
s
ter
in
g
b
ased
o
n
s
tu
d
en
t
b
eh
a
v
io
r
f
o
llo
wed
b
y
a
d
ec
is
io
n
tr
ee
.
Sti
ll,
r
esu
lts
w
er
e
s
p
ec
if
ic
to
th
e
ex
is
tin
g
e
-
lear
n
in
g
p
latf
o
r
m
with
lim
ited
g
en
er
aliza
b
ilit
y
an
d
d
i
d
n
o
t
co
n
s
id
er
VR
o
r
i
m
m
er
s
iv
e
lear
n
in
g
.
Vital
et
a
l.
[
8
]
an
aly
ze
d
s
tu
d
en
t
p
e
r
f
o
r
m
an
ce
u
s
in
g
s
tatis
tical
m
eth
o
d
s
co
m
b
in
ed
with
K
-
Me
an
s
an
d
h
ier
ar
ch
ical
clu
s
ter
in
g
,
s
tu
d
y
i
n
g
f
ac
to
r
s
s
u
ch
as
f
am
ily
b
a
ck
g
r
o
u
n
d
,
p
er
s
o
n
al
p
r
o
f
ile,
a
n
d
life
s
ty
le
h
ab
its
,
with
clu
s
ter
in
g
h
el
p
in
g
t
o
p
r
ed
ict
p
ass
/f
ail
o
u
tco
m
es
a
n
d
u
n
d
er
s
tan
d
u
n
d
e
r
ly
in
g
ca
u
s
es.
Oth
er
s
tu
d
ies
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J E
lec
&
C
o
m
p
E
n
g
I
SS
N:
2088
-
8
7
0
8
S
tu
d
en
ts
p
erfo
r
ma
n
ce
clu
s
teri
n
g
fo
r
fu
tu
r
e
p
ers
o
n
a
liz
ed
in
…
(
Gh
a
lia
Md
a
g
h
r
i A
la
o
u
i
)
299
co
m
b
in
ed
clu
s
ter
in
g
with
ad
d
i
tio
n
al
tech
n
iq
u
es:
Pra
b
h
a
an
d
Priy
aa
[
2
0
]
ap
p
lied
f
u
zz
y
K
-
m
ed
o
id
s
b
u
t
lack
ed
ex
ter
n
al
v
alid
atio
n
o
r
s
u
f
f
er
e
d
h
ig
h
co
m
p
u
tatio
n
al
c
o
s
ts
,
Haf
d
i
an
d
E
l
Kaf
h
ali
[
2
1
]
e
x
p
lo
r
ed
p
r
ed
ictiv
e
m
o
d
elin
g
with
s
m
all
d
atasets
,
an
d
Xu
et
a
l
.
[
2
2
]
in
v
esti
g
ated
p
er
f
o
r
m
an
ce
ev
o
lu
tio
n
b
u
t
f
ac
ed
s
en
s
itiv
ity
is
s
u
es.
Fo
r
d
r
o
p
o
u
t
p
r
ed
ictio
n
in
m
ass
iv
e
o
p
en
o
n
lin
e
co
u
r
s
e
(
MO
OC
)
v
ia
d
ee
p
lear
n
in
g
[
2
3
]
,
clu
s
ter
s
wer
e
n
o
t
ex
p
licitly
g
e
n
er
ated
,
wh
ile
Sh
ar
if
an
d
Atif
[
2
4
]
em
p
h
asized
th
e
b
en
e
f
its
o
f
p
er
s
o
n
alize
d
f
ee
d
b
ac
k
d
esp
ite
ch
allen
g
es
r
elate
d
to
p
r
iv
ac
y
an
d
co
n
te
x
tu
al
s
p
ec
if
icity
.
As
th
e
s
co
p
e
o
f
clu
s
ter
in
g
ex
ten
d
s
b
ey
o
n
d
p
er
f
o
r
m
an
ce
to
in
clu
d
e
p
e
r
s
o
n
aliza
tio
n
,
th
e
in
teg
r
atio
n
o
f
v
ir
tu
al
r
ea
lity
em
er
g
es
a
s
a
p
r
o
m
is
in
g
y
et
u
n
d
er
e
x
p
lo
r
e
d
ar
ea
.
2
.
2
.
Virt
ua
l r
e
a
lity
,
perso
na
lized
lea
rning
,
a
nd
a
da
ptiv
e
s
t
ud
ent
m
o
delin
g
VR
tech
n
o
lo
g
ies
ar
e
r
esh
ap
in
g
lear
n
in
g
b
y
p
r
o
v
id
in
g
in
ter
ac
tiv
e
s
im
u
latio
n
s
an
d
p
er
s
o
n
alize
d
co
n
ten
t
t
h
at
en
h
an
ce
en
g
a
g
e
m
en
t
an
d
o
u
tco
m
es
[
2
5
]
,
[
2
6
]
.
Mo
s
t
s
tu
d
ies
f
o
cu
s
o
n
s
cien
ce
an
d
m
ath
e
m
atics,
th
o
u
g
h
th
e
s
o
cial
s
cien
ce
s
also
ad
o
p
t
VR
f
o
r
ed
u
ca
tio
n
al
p
u
r
p
o
s
es.
W
h
ile
v
is
u
al
elem
en
ts
d
o
m
in
ate,
im
m
er
s
iv
e
in
ter
ac
tiv
ity
r
em
ai
n
s
lim
ited
,
h
ig
h
lig
h
tin
g
th
e
n
e
ed
f
o
r
f
u
r
th
er
r
esear
ch
.
Featu
r
es
s
u
ch
as
p
r
esen
ce
,
au
to
n
o
m
y
,
a
n
d
au
th
e
n
tic
task
s
s
u
p
p
o
r
t
lear
n
in
g
with
in
c
o
n
s
tr
u
ctiv
is
t
f
r
am
ewo
r
k
s
,
b
u
t
lo
n
g
itu
d
in
al
s
tu
d
ies
ar
e
n
ee
d
ed
to
ass
ess
k
n
o
wled
g
e
r
e
ten
tio
n
.
R
esear
ch
o
n
VR
-
b
ased
in
d
iv
i
d
u
alize
d
lear
n
in
g
a
n
d
s
tu
d
en
t
clu
s
ter
in
g
is
s
till
lim
ited
.
Per
s
o
n
alize
d
lear
n
in
g
r
eq
u
ir
es
s
o
p
h
is
ticated
p
r
o
f
ilin
g
to
ad
ap
t
c
o
n
ten
t,
p
ac
in
g
,
an
d
in
s
tr
u
ctio
n
al
s
tr
ateg
ies.
T
r
ad
itio
n
al
m
eth
o
d
s
o
f
te
n
r
ely
o
n
ass
es
s
m
en
ts
o
r
b
eh
a
v
io
r
al
tr
ac
k
in
g
,
wh
er
ea
s
AI
-
d
r
iv
en
ap
p
r
o
a
ch
es
en
ab
le
m
o
r
e
d
y
n
am
ic
lear
n
e
r
m
o
d
elin
g
.
Ad
ap
tiv
e
lear
n
in
g
tech
n
o
lo
g
ies,
b
o
o
s
ted
b
y
AI
an
d
th
e
s
u
r
g
e
in
d
ig
ital
ed
u
ca
tio
n
d
u
r
in
g
t
h
e
C
OVI
D
-
1
9
p
an
d
e
m
ic,
h
av
e
tr
an
s
f
o
r
m
ed
p
er
s
o
n
aliza
tio
n
,
ac
ce
s
s
ib
ilit
y
,
an
d
ef
f
icien
cy
,
s
u
p
p
o
r
tin
g
s
tu
d
en
t
-
ce
n
ter
ed
lear
n
in
g
,
f
o
s
ter
in
g
in
f
o
r
m
ed
citizen
s
,
an
d
p
r
o
m
o
tin
g
s
u
s
tain
ab
le
d
ev
elo
p
m
e
n
t
[
2
7
]
.
C
lu
s
ter
in
g
tech
n
iq
u
es,
in
p
ar
ticu
lar
,
o
f
f
er
p
r
o
m
is
in
g
av
e
n
u
es
to
g
en
er
ate
ac
tio
n
ab
le
lear
n
er
p
r
o
f
iles
,
b
u
t
o
p
er
atio
n
alizin
g
th
e
m
in
to
m
ea
n
in
g
f
u
l
s
tr
ateg
ies
with
in
i
m
m
er
s
iv
e
VR
en
v
ir
o
n
m
en
ts
r
em
ain
s
ch
allen
g
in
g
an
d
ca
lls
f
o
r
f
u
r
th
e
r
in
ter
d
is
ci
p
lin
ar
y
r
esear
c
h
.
Ad
ap
tiv
e
lear
n
in
g
p
latf
o
r
m
s
th
at
d
y
n
am
ically
a
d
ju
s
t
to
in
d
iv
id
u
al
lear
n
e
r
s
,
o
f
ten
th
r
o
u
g
h
m
u
lti
-
ag
en
t
s
y
s
tem
s
,
d
ep
en
d
o
n
co
m
p
r
eh
en
s
iv
e
s
tu
d
en
t
m
o
d
els
r
ep
r
esen
tin
g
p
r
ef
er
en
ce
s
,
en
g
ag
em
en
t
lev
els,
an
d
p
er
f
o
r
m
an
ce
p
atter
n
s
.
R
ec
en
t
ad
v
an
ce
s
in
AI
,
p
a
r
ticu
lar
ly
lar
g
e
lan
g
u
ag
e
m
o
d
els
(
L
L
Ms)
,
h
av
e
en
a
b
le
d
ag
en
tic
wo
r
k
f
lo
ws
(
AW
s
)
an
d
f
r
am
ewo
r
k
s
lik
e
Ag
en
t
4
E
D
U,
wh
ich
s
u
p
p
o
r
t
co
m
p
lex
ed
u
ca
tio
n
al
task
s
an
d
m
u
lti
-
ag
en
t
c
o
llab
o
r
atio
n
,
f
u
r
th
er
en
h
an
cin
g
a
d
ap
tiv
e
an
d
p
er
s
o
n
alize
d
lea
r
n
in
g
ex
p
er
ie
n
ce
s
[
2
8
]
.
Desp
ite
th
ese
ad
v
an
ce
m
e
n
ts
,
m
eth
o
d
o
lo
g
ical
in
c
o
n
s
is
ten
cies
in
clu
s
ter
in
g
s
tu
d
ies
co
n
tin
u
e
to
lim
it
b
r
o
ad
er
ap
p
licatio
n
an
d
r
ep
licatio
n
.
2
.
3
.
Adv
a
nced
m
a
chine le
a
rning
t
ec
hn
iq
ues
in o
t
her
do
m
a
ins
W
h
ile
m
o
s
t
clu
s
ter
in
g
an
d
p
r
ed
ictiv
e
m
o
d
elin
g
s
tu
d
ies
in
ed
u
ca
tio
n
r
em
ain
lim
ited
in
s
co
p
e,
ad
v
an
ce
s
in
o
th
e
r
f
ield
s
h
ig
h
l
ig
h
t
th
e
p
o
ten
tial
o
f
m
ac
h
in
e
lear
n
in
g
t
o
ca
p
t
u
r
e
c
o
m
p
lex
,
n
o
n
lin
ea
r
p
atter
n
s
.
T
h
ese
ac
h
iev
em
en
ts
,
alth
o
u
g
h
o
u
ts
id
e
th
e
ed
u
ca
tio
n
al
co
n
te
x
t,
p
r
o
v
i
d
e
m
eth
o
d
o
lo
g
ical
in
s
ig
h
ts
th
at
m
o
tiv
ate
o
u
r
ex
p
lo
r
atio
n
o
f
ad
v
an
ce
d
clu
s
ter
in
g
an
d
class
if
icatio
n
ap
p
r
o
ac
h
es
f
o
r
s
tu
d
en
t
p
r
o
f
ilin
g
in
VR
-
b
ased
p
er
s
o
n
alize
d
lear
n
i
n
g
.
R
ec
en
t
ad
v
an
ce
s
in
p
r
ed
ictiv
e
m
o
d
elin
g
—
n
e
u
r
al
n
etwo
r
k
s
,
Gau
s
s
ian
p
r
o
ce
s
s
r
eg
r
ess
io
n
(
GPR
)
,
g
r
ap
h
-
b
ased
,
a
n
d
en
s
em
b
le
m
eth
o
d
s
—
ef
f
ec
tiv
ely
ca
p
tu
r
e
c
o
m
p
lex
n
o
n
lin
ea
r
p
atter
n
s
,
m
o
tiv
atin
g
th
e
u
s
e
o
f
m
u
ltip
le
clu
s
ter
in
g
an
d
cla
s
s
if
icatio
n
ap
p
r
o
ac
h
es
f
o
r
an
aly
zin
g
s
tu
d
en
t
p
er
f
o
r
m
a
n
ce
in
VR
-
b
ased
p
er
s
o
n
alize
d
lear
n
in
g
.
Neu
r
al
n
etwo
r
k
s
(
NAR
-
NN)
f
o
r
ec
a
s
ted
th
er
m
al
co
al
tr
ad
in
g
v
o
l
u
m
es
(
2
0
1
6
–
2
0
2
0
)
with
m
in
im
al
er
r
o
r
u
p
to
th
e
9
9
.
2
7
3
t
h
q
u
a
n
tile
[
9
]
an
d
wee
k
ly
p
ea
n
u
t
o
il
p
r
ices
with
tr
ain
in
g
,
v
alid
atio
n
,
an
d
test
in
g
r
o
o
t
m
ea
n
s
q
u
ar
e
d
er
r
o
r
(
R
MSE
)
o
f
5
.
8
9
,
4
.
9
6
,
a
n
d
5
.
5
7
[
1
0
]
.
GPR
with
B
ay
esian
o
p
tim
izatio
n
ac
cu
r
ately
p
r
e
d
icted
d
aily
s
il
v
er
p
r
ices
o
v
er
1
3
y
ea
r
s
(
r
ela
tiv
e
R
MSE
0
.
2
2
5
7
%,
co
r
r
elat
io
n
9
9
.
9
6
7
%)
[
1
1
]
an
d
th
er
m
al
co
al
p
r
ices
(
r
elativ
e
R
MSE
0
.
4
2
1
0
%)
[
1
2
]
.
Gr
ap
h
ical
m
o
d
els,
in
clu
d
in
g
DAGs,
r
ev
ea
led
d
y
n
am
ic
in
ter
ac
tio
n
s
am
o
n
g
C
h
in
ese
p
r
o
p
er
t
y
in
d
ices
[
1
3
]
an
d
co
n
tem
p
o
r
a
n
eo
u
s
lin
k
ag
es
am
o
n
g
US
co
r
n
f
u
tu
r
es
an
d
ca
s
h
p
r
ices
[
1
4
]
.
E
n
s
em
b
le
an
d
c
o
m
p
o
s
ite
m
et
h
o
d
s
en
h
an
ce
d
r
o
b
u
s
tn
ess
,
wi
th
3
0
m
o
d
els
an
d
1
0
co
m
b
in
atio
n
s
r
ed
u
ci
n
g
er
r
o
r
s
in
d
aily
co
r
n
p
r
ices
[
1
5
]
an
d
5
1
m
o
d
els
with
4
1
en
s
em
b
le
v
ar
iatio
n
s
ac
h
iev
i
n
g
s
tr
o
n
g
p
e
r
f
o
r
m
an
ce
f
o
r
th
e
C
h
in
ese
s
to
ck
in
d
e
x
[
1
6
]
.
T
h
es
e
r
esu
lts
d
em
o
n
s
tr
ate
th
e
p
o
t
en
tial
o
f
a
d
v
an
ce
d
m
o
d
elin
g
tec
h
n
iq
u
es to
i
d
en
tif
y
n
o
n
li
n
ea
r
p
atter
n
s
an
d
g
u
id
e
ad
ap
tiv
e,
p
er
s
o
n
alize
d
VR
lear
n
in
g
.
2
.
4
.
M
et
ho
do
lo
g
ic
a
l g
a
ps
a
n
d da
t
a
s
et
lim
it
a
t
io
ns
in educa
t
io
na
l c
lus
t
er
ing
re
s
ea
rc
h
A
r
ev
iew
o
f
cl
u
s
ter
in
g
ap
p
licatio
n
s
in
ed
u
ca
tio
n
al
co
n
te
x
ts
r
ev
ea
ls
im
p
o
r
tan
t
c
h
allen
g
es
r
e
lated
b
o
th
to
ev
alu
atio
n
p
r
ac
tices
an
d
to
th
e
d
atasets
em
p
lo
y
e
d
.
I
n
ter
m
s
o
f
v
alid
atio
n
,
s
tu
d
ies
r
ely
o
n
d
i
v
er
s
e
m
etr
ics
—
r
an
g
in
g
f
r
o
m
i
n
ter
n
al
co
h
esio
n
in
d
icato
r
s
to
ex
ter
n
al
class
i
f
icatio
n
-
b
ased
v
alid
atio
n
s
—
m
ak
in
g
it
d
if
f
icu
lt
t
o
co
m
p
ar
e
f
in
d
in
g
s
o
r
r
ep
r
o
d
u
c
e
m
eth
o
d
o
lo
g
ies.
T
ab
le
1
p
r
e
s
en
ts
a
co
m
p
ar
ativ
e
s
u
m
m
ar
y
o
f
th
e
ev
alu
atio
n
tech
n
iq
u
es u
s
ed
in
k
ey
s
tu
d
ies
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8
7
0
8
I
n
t J E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
1
6
,
No
.
1
,
Feb
r
u
ar
y
20
2
6
:
2
9
7
-
310
300
T
h
is
co
m
p
ar
is
o
n
h
ig
h
lig
h
ts
k
ey
g
ap
s
in
e
d
u
ca
tio
n
al
clu
s
ter
in
g
r
esear
c
h
:
i
)
in
co
n
s
is
ten
t
v
alid
atio
n
m
etr
ics
lim
itin
g
r
ep
r
o
d
u
cib
ilit
y
,
ii
)
lack
o
f
ex
ter
n
al
v
alid
atio
n
v
ia
d
o
wn
s
tr
ea
m
task
s
,
iii
)
a
b
s
en
ce
o
f
s
tatis
tical
s
ig
n
if
ican
ce
test
in
g
f
o
r
alg
o
r
i
th
m
co
m
p
ar
is
o
n
s
,
a
n
d
iv
)
in
s
u
f
f
icien
t
atten
tio
n
to
ed
u
ca
tio
n
al
in
ter
p
r
etab
ilit
y
an
d
p
r
ac
tical
ap
p
licab
ilit
y
.
E
x
is
tin
g
s
tu
d
ies
also
r
ely
o
n
l
im
ited
,
o
f
te
n
o
n
e
-
d
im
e
n
s
io
n
a
l
d
atasets
f
r
o
m
e
-
lear
n
in
g
p
latf
o
r
m
s
,
f
o
cu
s
in
g
o
n
s
in
g
le
s
u
b
jects
o
r
n
ar
r
o
w
in
d
icato
r
s
r
ath
er
th
an
ca
p
tu
r
in
g
m
u
ltifa
ce
ted
s
tu
d
en
t
co
m
p
eten
cies.
Fu
r
th
er
m
o
r
e,
f
ew
wo
r
k
s
s
y
s
tem
atica
lly
co
m
p
ar
e
m
u
ltip
le
clu
s
ter
in
g
alg
o
r
ith
m
s
,
an
d
th
e
lack
o
f
s
tan
d
ar
d
ize
d
ev
al
u
atio
n
f
r
a
m
ewo
r
k
s
r
estricts
r
ep
licab
ilit
y
,
h
in
d
e
r
in
g
t
h
e
d
e
v
elo
p
m
en
t
o
f
r
o
b
u
s
t
b
est p
r
ac
tices.
T
ab
le
1
.
E
v
alu
atio
n
m
et
r
ics co
m
p
ar
is
o
n
ac
r
o
s
s
ed
u
ca
tio
n
al
c
lu
s
ter
in
g
s
tu
d
ies
S
t
u
d
y
I
n
t
e
r
n
a
l
m
e
t
r
i
c
s
Ex
t
e
r
n
a
l
v
a
l
i
d
a
t
i
o
n
S
t
a
t
i
st
i
c
a
l
t
e
s
t
i
n
g
Ed
u
c
a
t
i
o
n
a
l
i
n
t
e
r
p
r
e
t
a
t
i
o
n
G
o
v
i
n
d
a
s
a
m
y
a
n
d
V
e
l
m
u
r
u
g
a
n
[
2
9
]
N
M
I
,
P
u
r
i
t
y
N
o
t
u
se
d
N
o
t
u
se
d
Li
mi
t
e
d
N
a
v
a
r
r
o
e
t
a
l
.
[
5
]
S
i
l
h
o
u
e
t
t
e
,
D
B
I
n
d
e
x
N
o
t
u
se
d
N
o
t
u
se
d
Li
mi
t
e
d
V
i
t
a
l
e
t
a
l
.
[
8
]
V
i
su
a
l
i
n
s
p
e
c
t
i
o
n
C
l
a
s
si
f
i
c
a
t
i
o
n
a
c
c
u
r
a
c
y
N
o
t
u
se
d
Li
mi
t
e
d
K
r
i
ž
a
n
i
ć
[
7
]
N
o
t
r
e
p
o
r
t
e
d
D
e
c
i
s
i
o
n
t
r
e
e
v
a
l
i
d
a
t
i
o
n
N
o
t
u
se
d
Li
mi
t
e
d
Th
i
s
S
t
u
d
y
D
B
I
n
d
e
x
,
C
H
I
n
d
e
x
C
l
a
s
si
f
i
c
a
t
i
o
n
a
c
c
u
r
a
c
y
N
o
t
u
se
d
Li
mi
t
e
d
2
.
5
.
Resea
rc
h po
s
it
io
nin
g
a
nd
co
ntr
ibu
t
io
n
T
h
is
r
esear
ch
ad
d
r
ess
es
s
ev
er
al
cr
itical
g
ap
s
in
ed
u
ca
tio
n
a
l
clu
s
ter
in
g
liter
atu
r
e.
Pre
v
io
u
s
s
tu
d
ies
o
f
ten
lack
s
y
s
tem
atic
co
m
p
ar
i
s
o
n
s
b
etwe
en
m
u
ltip
le
clu
s
ter
in
g
alg
o
r
ith
m
s
with
r
ig
o
r
o
u
s
s
tatis
tical
v
alid
atio
n
,
f
o
cu
s
p
r
im
ar
ily
o
n
o
n
e
-
d
im
e
n
s
io
n
al
e
-
lear
n
in
g
d
ata,
a
n
d
p
r
o
v
id
e
lim
ited
co
n
ce
p
t
u
al
f
r
am
e
wo
r
k
s
f
o
r
ap
p
ly
in
g
clu
s
ter
in
g
r
esu
lts
in
p
er
s
o
n
ali
ze
d
lear
n
in
g
s
y
s
tem
s
,
p
a
r
ticu
lar
ly
in
v
ir
tu
al
r
ea
lity
en
v
ir
o
n
m
en
ts
.
Mo
r
e
o
v
er
,
ex
ter
n
al
v
alid
atio
n
o
f
clu
s
ter
s
th
r
o
u
g
h
co
n
cr
ete
ed
u
ca
tio
n
al
task
s
is
f
r
eq
u
en
tly
in
s
u
f
f
icie
n
t,
r
aisi
n
g
co
n
ce
r
n
s
ab
o
u
t th
e
p
r
ac
tical
ap
p
licab
ilit
y
o
f
th
e
r
esu
lts
.
T
o
o
v
er
c
o
m
e
th
ese
lim
itatio
n
s
,
th
is
s
tu
d
y
p
r
esen
ts
a
co
m
p
r
eh
en
s
iv
e
co
m
p
ar
is
o
n
o
f
f
iv
e
clu
s
ter
in
g
alg
o
r
ith
m
s
ap
p
lied
to
a
m
u
ltid
im
en
s
io
n
al
d
ataset
en
co
m
p
ass
in
g
m
ath
em
atics,
r
ea
d
in
g
,
an
d
wr
itin
g
,
en
ab
lin
g
a
d
etailed
an
aly
s
is
o
f
s
tu
d
en
t
p
r
o
f
iles
.
I
t
d
ev
elo
p
s
a
r
o
b
u
s
t
ass
ess
m
en
t
f
r
am
ewo
r
k
co
m
b
in
i
n
g
in
ter
n
al
in
d
ices,
ex
ter
n
al
v
alid
atio
n
th
r
o
u
g
h
cl
ass
if
icatio
n
,
an
d
r
ig
o
r
o
u
s
s
tati
s
tical
test
s
.
I
n
ad
d
itio
n
,
it
in
tr
o
d
u
ce
s
a
co
n
ce
p
tu
al
f
r
am
ewo
r
k
f
o
r
in
teg
r
atin
g
th
ese
p
r
o
f
iles
in
to
p
e
r
s
o
n
alize
d
lear
n
in
g
en
v
i
r
o
n
m
e
n
ts
in
v
ir
tu
al
r
ea
lity
.
T
h
e
th
eo
r
etica
l
co
n
tr
ib
u
tio
n
s
d
em
o
n
s
tr
ate
th
e
s
u
p
er
io
r
ity
o
f
s
p
e
ctr
u
m
clu
s
ter
in
g
o
v
er
co
n
v
en
t
io
n
al
m
eth
o
d
s
lik
e
K
-
Me
an
s
,
p
r
o
m
p
tin
g
a
r
ec
o
n
s
id
er
atio
n
o
f
a
n
aly
tical
ap
p
r
o
ac
h
es
to
ca
p
tu
r
e
th
e
co
m
p
le
x
ity
o
f
e
d
u
ca
tio
n
al
d
ata.
Pra
ctica
lly
,
th
e
s
tu
d
y
p
r
o
v
id
es
a
co
n
cr
ete
r
o
ad
m
ap
f
o
r
p
e
r
s
o
n
aliza
tio
n
in
im
m
er
s
iv
e
v
ir
tu
al
r
ea
lity
en
v
ir
o
n
m
en
ts
,
f
ac
ilit
atin
g
th
e
in
teg
r
atio
n
o
f
d
ata
a
n
aly
s
is
in
t
o
ed
u
ca
tio
n
al
s
y
s
tem
s
an
d
lay
in
g
th
e
f
o
u
n
d
atio
n
f
o
r
th
e
n
ex
t g
e
n
er
atio
n
o
f
in
tel
lig
en
t a
n
d
a
d
ap
tiv
e
lear
n
in
g
s
y
s
tem
s
.
3.
M
E
T
H
O
DO
L
O
G
Y
3
.
1
.
Resea
rc
h desi
g
n a
n
d o
v
er
a
ll a
rc
hite
ct
ure
T
h
is
s
tu
d
y
em
p
lo
y
s
a
h
y
b
r
id
m
ac
h
in
e
-
lear
n
in
g
f
r
am
ew
o
r
k
th
at
co
m
b
in
es
u
n
s
u
p
er
v
is
ed
an
d
s
u
p
er
v
is
ed
tech
n
iq
u
es
to
d
ev
e
lo
p
p
er
s
o
n
alize
d
lear
n
in
g
p
ath
way
s
f
o
r
s
tu
d
en
ts
'
f
u
tu
r
e
u
s
e.
Fig
u
r
e
1
s
h
o
ws
th
e
f
r
am
ewo
r
k
f
lo
w
f
r
o
m
d
ata
in
p
u
t
th
r
o
u
g
h
cl
u
s
ter
in
g
a
n
d
class
if
icatio
n
to
VR
-
b
ased
ad
ap
tiv
e
lear
n
in
g
in
teg
r
atio
n
.
T
h
e
p
r
o
p
o
s
ed
m
e
th
o
d
o
lo
g
y
is
s
tr
u
ctu
r
ed
in
t
o
t
h
r
ee
m
ain
p
h
ases
:
i
)
u
n
s
u
p
er
v
is
ed
clu
s
ter
in
g
to
u
n
co
v
e
r
n
atu
r
al
g
r
o
u
p
in
g
s
am
o
n
g
s
tu
d
en
ts
with
o
u
t
r
ely
i
n
g
o
n
p
r
ed
ef
in
ed
lab
els,
ii
)
s
u
p
er
v
is
ed
class
if
icatio
n
to
p
r
ed
ict
th
e
g
r
o
u
p
m
em
b
e
r
s
h
ip
o
f
n
ew
s
tu
d
e
n
ts
b
ased
o
n
th
eir
ac
ad
e
m
ic
p
er
f
o
r
m
an
ce
,
an
d
iii
)
th
e
co
n
ce
p
tu
al
in
te
g
r
atio
n
o
f
ad
a
p
tiv
e
lear
n
in
g
r
o
u
tes in
to
a
VR
p
latf
o
r
m
.
T
h
e
in
n
o
v
ativ
e
asp
ec
t
o
f
t
h
is
ap
p
r
o
ac
h
lies
in
tr
an
s
f
o
r
m
i
n
g
u
n
lab
eled
clu
s
ter
in
g
o
u
tp
u
ts
in
to
lab
eled
tar
g
ets
f
o
r
s
u
p
er
v
is
ed
lear
n
i
n
g
,
e
n
ab
lin
g
th
e
co
n
s
tr
u
ctio
n
o
f
p
r
e
d
ictiv
e
m
o
d
els
b
ased
o
n
em
p
ir
ically
d
is
co
v
er
ed
p
atter
n
s
.
Un
lik
e
tr
ad
itio
n
al
class
if
icatio
n
s
y
s
tem
s
,
th
is
m
eth
o
d
o
lo
g
y
f
ir
s
t
d
ete
cts
laten
t
s
tr
u
ctu
r
es
in
s
tu
d
en
t
d
ata
—
s
p
ec
if
ically
,
ac
ad
em
ic
p
er
f
o
r
m
a
n
ce
in
d
icato
r
s
s
u
ch
as
m
ath
,
r
ea
d
i
n
g
,
a
n
d
wr
itin
g
s
co
r
es
—
u
s
in
g
clu
s
ter
in
g
alg
o
r
ith
m
s
(
K
-
Me
an
s
,
h
ier
ar
c
h
ical
clu
s
ter
in
g
(
HC
)
,
Gau
s
s
ian
m
ix
tu
r
e
m
o
d
el
(
GM
M)
,
s
p
ec
tr
al
clu
s
ter
in
g
,
an
d
K
-
M
ed
o
id
s
)
.
T
h
e
clu
s
ter
in
g
p
er
f
o
r
m
an
ce
is
ev
alu
ate
d
th
r
o
u
g
h
in
ter
n
al
v
alid
atio
n
in
d
ices to
d
eter
m
in
e
t
h
e
m
o
s
t
s
u
itab
le
f
ea
tu
r
es a
n
d
alg
o
r
ith
m
s
.
Nex
t,
s
u
p
er
v
is
ed
class
if
icatio
n
m
o
d
els,
s
u
ch
as
d
ec
is
io
n
tr
ee
s
(
DT
s
)
,
s
u
p
p
o
r
t
v
ec
t
o
r
m
ac
h
in
e
(
SVM
)
,
lo
g
is
tic
r
eg
r
ess
io
n
(
L
R
)
,
K
-
n
ea
r
est
n
eig
h
b
o
r
s
(
K
-
NN
)
,
an
d
r
an
d
o
m
f
o
r
est
(
R
F)
,
a
r
e
tr
ain
ed
to
p
r
ed
ict
th
e
clu
s
ter
m
em
b
er
s
h
ip
o
f
n
e
w
s
tu
d
en
ts
.
T
h
e
b
est
-
p
er
f
o
r
m
i
n
g
m
o
d
el
is
s
elec
ted
b
ased
o
n
ac
cu
r
ac
y
s
co
r
es
an
d
u
s
ed
to
s
im
u
late
s
tu
d
en
t
g
r
o
u
p
in
g
f
o
r
p
er
s
o
n
alize
d
i
n
ter
v
en
tio
n
.
T
h
e
VR
co
m
p
o
n
en
t
is
th
e
n
ex
t
s
tag
e
o
f
th
is
r
esear
ch
,
ev
en
i
f
it
h
as
n
o
t
b
ee
n
u
s
ed
y
et.
W
ith
ea
ch
clu
s
ter
ac
tin
g
as
th
e
b
asis
f
o
r
tr
i
g
g
er
in
g
r
ea
l
-
tim
e,
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J E
lec
&
C
o
m
p
E
n
g
I
SS
N:
2088
-
8
7
0
8
S
tu
d
en
ts
p
erfo
r
ma
n
ce
clu
s
teri
n
g
fo
r
fu
tu
r
e
p
ers
o
n
a
liz
ed
in
…
(
Gh
a
lia
Md
a
g
h
r
i A
la
o
u
i
)
301
p
r
o
f
ile
-
d
r
iv
en
ad
ju
s
tm
en
ts
in
im
m
er
s
iv
e
lear
n
in
g
en
v
ir
o
n
m
en
ts
,
th
e
l
o
n
g
-
te
r
m
g
o
al
is
to
in
c
o
r
p
o
r
ate
t
h
e
d
etec
ted
s
tu
d
en
t
p
r
o
f
iles
in
to
a
m
u
lti
-
ag
en
t
VR
-
b
ased
ed
u
ca
tio
n
al
s
y
s
tem
.
Fig
u
r
e
1
.
Hy
b
r
id
f
r
am
ewo
r
k
c
o
m
b
in
in
g
clu
s
ter
in
g
,
class
if
icatio
n
,
an
d
VR
-
b
ased
p
er
s
o
n
aliz
atio
n
3
.
2
.
Da
t
a
s
et
des
cr
iptio
n a
nd
prepa
ra
t
io
n
T
h
is
s
tu
d
y
em
p
lo
y
s
a
d
ataset
co
n
s
is
tin
g
o
f
ac
ad
em
ic
p
er
f
o
r
m
an
ce
m
etr
ics
f
r
o
m
1
,
0
0
0
s
tu
d
en
ts
in
th
r
ee
co
r
e
s
u
b
jects: m
ath
em
atics,
wr
itin
g
,
an
d
r
ea
d
in
g
.
T
h
es
e
s
u
b
jects we
r
e
s
p
ec
if
ically
s
e
lecte
d
b
ec
au
s
e
th
ey
p
r
o
v
id
e
a
co
m
p
r
eh
en
s
iv
e
u
n
d
er
s
tan
d
in
g
o
f
s
tu
d
en
ts
’
ac
a
d
em
ic
ab
ilit
ies
an
d
r
ep
r
esen
t
ess
en
tial
ac
ad
em
ic
co
m
p
eten
cies.
T
h
e
d
ataset,
o
b
tain
ed
f
r
o
m
e
d
u
ca
tio
n
al
in
s
titu
tio
n
s
,
in
clu
d
es
s
tan
d
ar
d
ize
d
t
est
r
esu
lts
f
o
r
ea
ch
s
u
b
ject
[
3
0
]
.
An
o
v
er
v
iew
o
f
t
h
e
d
ataset
is
p
r
esen
ted
in
T
ab
l
e
2
.
T
ab
le
2
.
Aca
d
em
ic
p
er
f
o
r
m
an
ce
s
co
r
es
S
t
u
d
e
n
t
I
D
M
a
t
h
sc
o
r
e
W
r
i
t
i
n
g
sc
o
r
e
R
e
a
d
i
n
g
s
c
o
r
e
0
72
72
74
1
69
90
88
2
47
57
44
.
.
.
.
.
.
….
.
.
….
.
….
.
9
9
9
88
99
95
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8
7
0
8
I
n
t J E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
1
6
,
No
.
1
,
Feb
r
u
ar
y
20
2
6
:
2
9
7
-
310
302
T
o
en
s
u
r
e
d
ata
q
u
ality
an
d
p
r
ep
ar
e
t
h
e
d
ataset
f
o
r
s
u
b
s
eq
u
en
t
clu
s
ter
in
g
an
d
class
if
icatio
n
task
s
,
s
ev
er
al
p
r
ep
r
o
ce
s
s
in
g
s
tep
s
w
er
e
ap
p
lied
.
Miss
in
g
v
alu
es
i
n
n
u
m
er
ical
f
ea
tu
r
es
wer
e
tr
e
ated
u
s
in
g
m
ed
ian
im
p
u
tatio
n
,
wh
ile
o
u
tlier
s
w
er
e
id
en
tifie
d
a
n
d
h
an
d
led
t
h
r
o
u
g
h
th
e
i
n
ter
q
u
a
r
tile
r
an
g
e
(
I
QR
)
m
eth
o
d
.
Dis
tr
ib
u
tio
n
an
aly
s
is
an
d
s
t
atis
tical
s
u
m
m
ar
ies
wer
e
co
n
d
u
cted
to
v
er
if
y
d
ata
in
te
g
r
ity
,
f
o
llo
wed
b
y
s
tan
d
ar
d
izatio
n
u
s
in
g
th
e
z
-
s
co
r
e
tr
an
s
f
o
r
m
atio
n
,
d
ef
in
e
d
as z
=
(
x
−
μ
)
/ σ
[
3
1
]
.
T
h
is
s
ca
li
n
g
s
tep
en
s
u
r
es th
at
all
th
r
ee
ac
ad
em
ic
s
u
b
jects
co
n
tr
ib
u
te
eq
u
ally
to
d
is
tan
ce
-
b
ased
clu
s
ter
in
g
alg
o
r
ith
m
s
.
A
s
u
m
m
ar
y
o
f
th
e
p
r
ep
r
o
ce
s
s
in
g
p
ip
elin
e
is
p
r
o
v
id
ed
in
T
ab
le
3
.
T
ab
le
3
.
Ov
e
r
v
iew
o
f
t
h
e
d
ata
p
r
ep
r
o
ce
s
s
in
g
p
ip
elin
e
S
t
e
p
D
e
scri
p
t
i
o
n
M
i
ss
i
n
g
V
a
l
u
e
H
a
n
d
l
i
n
g
M
e
d
i
a
n
i
m
p
u
t
a
t
i
o
n
a
p
p
l
i
e
d
t
o
n
u
m
e
r
i
c
a
l
f
e
a
t
u
r
e
s
O
u
t
l
i
e
r
D
e
t
e
c
t
i
o
n
Th
e
i
n
t
e
r
q
u
a
r
t
i
l
e
r
a
n
g
e
(
I
Q
R
)
me
t
h
o
d
i
s
u
se
d
t
o
i
d
e
n
t
i
f
y
a
n
d
t
r
e
a
t
o
u
t
l
i
e
r
s
D
a
t
a
Q
u
a
l
i
t
y
v
e
r
i
f
i
c
a
t
i
o
n
S
t
a
n
d
a
r
d
i
z
a
t
i
o
n
Ju
st
i
f
i
c
a
t
i
o
n
f
o
r
S
c
a
l
i
n
g
D
i
st
r
i
b
u
t
i
o
n
a
n
a
l
y
si
s
a
n
d
st
a
t
i
st
i
c
a
l
s
u
mm
a
r
i
e
s
a
r
e
u
s
e
d
t
o
e
n
s
u
r
e
d
a
t
a
i
n
t
e
g
r
i
t
y
F
e
a
t
u
r
e
s
sca
l
e
d
u
si
n
g
S
t
a
n
d
a
r
d
S
c
a
l
e
r
:
z
=
(
x
-
μ)
/
σ
En
s
u
r
e
s
e
q
u
a
l
c
o
n
t
r
i
b
u
t
i
o
n
o
f
a
l
l
a
c
a
d
e
mi
c
su
b
j
e
c
t
s
i
n
d
i
st
a
n
c
e
-
b
a
s
e
d
c
l
u
s
t
e
r
i
n
g
a
l
g
o
r
i
t
h
ms
3
.
3
.
Uns
up
er
v
is
ed
clu
s
t
er
in
g
a
pp
ro
a
ch
T
h
is
s
tu
d
y
em
p
lo
y
ed
f
iv
e
cl
u
s
ter
in
g
alg
o
r
ith
m
s
s
elec
ted
f
o
r
th
eir
co
m
p
lem
e
n
tar
y
s
tr
e
n
g
th
s
an
d
s
u
itab
ilit
y
f
o
r
ed
u
ca
tio
n
al
d
ata
an
aly
s
is
.
T
h
e
K
-
Me
an
s
alg
o
r
i
th
m
p
ar
titi
o
n
s
d
ata
in
to
k
well
-
s
ep
ar
ated
clu
s
ter
s
b
y
m
in
im
izin
g
th
e
with
in
-
clu
s
ter
s
u
m
o
f
s
q
u
a
r
ed
d
is
tan
ce
s
,
m
ak
in
g
it
ef
f
ec
tiv
e
f
o
r
co
n
tin
u
o
u
s
n
u
m
er
ical
v
ar
iab
les
an
d
s
u
itab
le
f
o
r
ca
teg
o
r
izin
g
s
tu
d
en
ts
b
y
p
er
f
o
r
m
an
ce
.
T
o
en
s
u
r
e
r
ep
r
o
d
u
ci
b
ilit
y
an
d
ef
f
icien
t
co
n
v
er
g
en
ce
,
s
cik
it
-
lear
n
’
s
im
p
lem
en
tatio
n
was
u
s
ed
with
in
it
=
'
r
an
d
o
m
'
an
d
ma
x_
iter
=
3
0
0
.
As
h
ig
h
lig
h
ted
b
y
Alza
h
r
an
i
et
a
l.
[
3
2
]
,
f
ea
t
u
r
e
s
tan
d
ar
d
izatio
n
with
z
-
s
co
r
e
tr
an
s
f
o
r
m
atio
n
was
ess
en
tial
to
av
o
id
b
ias
wh
en
v
ar
iab
les
h
ad
d
if
f
er
en
t
s
ca
les.
Hier
ar
ch
ical
c
l
u
s
ter
in
g
(
HC
)
was
also
ap
p
lied
to
e
x
p
lo
r
e
s
u
b
g
r
o
u
p
s
tr
u
ctu
r
es,
as
it
b
u
ild
s
a
tr
ee
-
lik
e
h
ier
ar
ch
y
o
f
clu
s
ter
s
an
d
r
e
v
ea
ls
s
tr
atif
ied
lin
k
s
an
d
n
ested
g
r
o
u
p
s
with
in
s
tu
d
en
t
p
er
f
o
r
m
an
ce
d
ata.
T
h
e
W
ar
d
lin
k
ag
e
cr
iter
io
n
was
ch
o
s
e
n
f
o
r
ag
g
lo
m
er
ativ
e
clu
s
ter
in
g
,
as
it
m
in
im
izes
v
ar
ian
ce
with
in
ea
ch
clu
s
ter
a
n
d
ten
d
s
to
p
r
o
d
u
ce
b
alan
ce
d
,
in
ter
p
r
etab
le
g
r
o
u
p
s
[
3
3
]
.
GM
M
was
in
clu
d
ed
to
r
ep
r
esen
t
th
e
d
ata
as
a
m
ix
tu
r
e
o
f
Gau
s
s
ian
d
is
tr
ib
u
tio
n
s
,
p
r
o
v
id
i
n
g
p
r
o
b
a
b
ilis
tic
clu
s
ter
m
em
b
er
s
h
ip
s
an
d
ca
p
tu
r
in
g
o
v
er
lap
p
i
n
g
clu
s
ter
s
.
T
h
is
was
p
ar
ticu
lar
ly
v
alu
ab
le
f
o
r
s
tu
d
en
t
p
er
f
o
r
m
an
ce
d
ata,
wh
er
e
in
d
iv
id
u
als
m
ay
s
im
u
ltan
eo
u
s
ly
ex
h
ib
it
tr
aits
o
f
m
u
lti
p
le
ca
teg
o
r
ies.
T
h
e
E
x
p
ec
tatio
n
-
Ma
x
im
izatio
n
(
E
M)
alg
o
r
ith
m
with
f
u
ll
c
o
v
ar
ia
n
ce
m
atr
ices
was
u
s
ed
in
im
p
l
em
en
tatio
n
[
3
4
]
.
I
n
co
n
tr
ast,
K
-
Me
d
o
id
s
(
PAM)
was
ap
p
lied
f
o
r
its
r
o
b
u
s
tn
ess
to
n
o
is
e
a
n
d
o
u
tlier
s
,
as
it
s
elec
ts
ac
tu
al
d
ata
p
o
in
ts
—
m
ed
o
id
s
—
as
clu
s
ter
ce
n
ter
s
.
T
h
is
p
r
eser
v
ed
r
ep
r
esen
tativ
e
s
tu
d
en
t
p
r
o
f
iles
an
d
im
p
r
o
v
e
d
in
ter
p
r
etab
ilit
y
,
u
s
in
g
th
e
Ma
n
h
attan
d
is
tan
ce
m
etr
ic
f
o
r
s
im
ilar
ity
ca
lcu
latio
n
s
[
3
5
]
.
Sp
ec
tr
al
clu
s
ter
in
g
was
f
in
a
lly
em
p
lo
y
ed
to
u
n
co
v
er
s
u
b
tle
p
er
f
o
r
m
an
ce
p
atter
n
s
th
at
lin
ea
r
ap
p
r
o
ac
h
es
m
ig
h
t
o
v
e
r
lo
o
k
.
B
y
lev
er
ag
in
g
E
ig
e
n
d
ec
o
m
p
o
s
itio
n
o
f
a
s
im
ilar
ity
m
atr
ix
an
d
co
m
b
in
in
g
n
o
r
m
alize
d
s
p
ec
tr
al
clu
s
ter
in
g
with
a
k
-
n
ea
r
est
n
eig
h
b
o
r
s
g
r
ap
h
,
th
is
m
eth
o
d
ca
p
tu
r
e
d
co
m
p
lex
an
d
n
o
n
lin
ea
r
d
ata
s
tr
u
ctu
r
es
e
f
f
ec
tiv
ely
[
3
6
]
.
T
h
e
co
m
b
i
n
atio
n
o
f
th
ese
f
iv
e
alg
o
r
ith
m
s
e
n
s
u
r
ed
b
o
th
d
iv
e
r
s
ity
an
d
r
o
b
u
s
tn
ess
in
u
n
c
o
v
er
in
g
p
er
f
o
r
m
an
ce
-
b
ased
s
tu
d
en
t
g
r
o
u
p
i
n
g
s
.
T
h
e
im
p
lem
en
tatio
n
p
r
o
ce
d
u
r
e
f
o
llo
wed
a
s
y
s
tem
atic
p
ip
eli
n
e
to
en
s
u
r
e
r
eliab
le
r
esu
lts
.
T
h
e
o
p
tim
al
n
u
m
b
er
o
f
clu
s
ter
s
k
was
d
et
er
m
in
ed
b
y
v
a
r
y
in
g
k
f
r
o
m
2
to
1
0
,
with
ea
ch
alg
o
r
ith
m
ex
ec
u
ted
ac
r
o
s
s
3
0
r
an
d
o
m
in
itializatio
n
s
to
en
s
u
r
e
s
tab
ilit
y
,
as
r
ec
o
m
m
en
d
ed
in
[
3
7
]
.
C
lu
s
ter
q
u
ality
was
ass
es
s
ed
u
s
in
g
two
in
ter
n
al
v
alid
ati
o
n
i
n
d
ices:
th
e
Dav
ies
-
B
o
u
ld
in
in
d
ex
(
DB
)
,
wh
er
e
lo
wer
v
alu
es
in
d
icate
d
b
etter
s
ep
ar
atio
n
an
d
c
o
m
p
ac
tn
ess
[
3
8
]
,
an
d
th
e
C
alin
s
k
i
-
Har
ab
asz
in
d
ex
(
C
H)
,
wh
er
e
h
ig
h
er
v
alu
es
r
e
f
lecte
d
well
-
d
ef
in
e
d
a
n
d
d
is
tin
ct
clu
s
ter
s
[
3
9
]
.
T
o
f
u
r
th
er
s
tr
en
g
th
e
n
th
e
c
o
m
p
ar
is
o
n
,
p
ai
r
ed
t
-
test
s
wer
e
co
n
d
u
cted
to
ass
ess
th
e
s
tatis
t
ical
s
ig
n
if
ican
ce
o
f
p
er
f
o
r
m
an
ce
d
if
f
e
r
en
ce
s
b
etwe
en
alg
o
r
ith
m
s
.
T
h
e
alg
o
r
ith
m
th
a
t
ac
h
iev
ed
th
e
m
o
s
t
f
av
o
r
a
b
le
in
d
ex
s
co
r
es a
n
d
s
ta
tis
tically
s
ig
n
if
ican
t r
esu
lts
wa
s
id
en
tifie
d
as th
e
b
est
-
p
er
f
o
r
m
in
g
m
eth
o
d
[
4
0
]
.
3
.
4
.
Su
perv
is
ed
cla
s
s
if
ica
t
io
n m
et
ho
do
lo
g
y
T
h
e
o
p
tim
al
clu
s
ter
in
g
s
o
lu
tio
n
p
r
o
d
u
ce
s
ca
teg
o
r
ical
lab
els
f
o
r
ea
ch
s
tu
d
en
t,
tr
an
s
f
o
r
m
in
g
th
e
u
n
s
u
p
er
v
is
ed
lear
n
in
g
task
in
t
o
a
s
u
p
er
v
is
ed
class
if
icatio
n
p
r
o
b
lem
.
T
h
ese
la
b
els
s
er
v
e
as
tar
g
et
v
a
r
iab
les
f
o
r
tr
ain
in
g
p
r
ed
ictiv
e
m
o
d
els,
e
n
ab
lin
g
au
to
m
atic
ca
teg
o
r
izat
io
n
o
f
n
ew
s
tu
d
en
ts
a
n
d
p
r
o
v
id
in
g
an
i
n
d
ir
ec
t
m
ea
s
u
r
e
o
f
clu
s
ter
s
tab
ilit
y
th
r
o
u
g
h
class
if
icatio
n
ac
cu
r
ac
y
[
4
1
]
.
Fiv
e
s
u
p
er
v
is
ed
lear
n
i
n
g
alg
o
r
ith
m
s
wer
e
im
p
lem
en
ted
:
DT
,
SVM,
LR
,
K
-
NN,
an
d
RF
.
T
h
e
DT
co
n
s
tr
u
cts
r
u
le
-
b
ased
b
o
u
n
d
ar
ies
v
ia
r
ec
u
r
s
iv
e
p
ar
titi
o
n
in
g
an
d
was
co
n
f
ig
u
r
ed
with
a
m
a
x
im
u
m
d
ep
t
h
o
f
1
0
an
d
a
m
in
im
u
m
o
f
f
iv
e
s
a
m
p
les
p
er
leaf
[
4
2
]
.
SVM
f
in
d
s
th
e
o
p
tim
al
h
y
p
er
p
lan
e
th
at
m
ax
im
izes
class
m
ar
g
in
s
,
u
s
in
g
an
R
B
F
k
er
n
el
with
C
=
1
.
0
an
d
g
am
m
a
s
et
to
“scale”
[
4
3
]
.
LR
m
o
d
els
clas
s
p
r
o
b
ab
ilit
ies
u
s
in
g
th
e
lo
g
is
tic
f
u
n
ctio
n
,
tr
ain
ed
with
L
2
r
eg
u
lar
izatio
n
f
o
r
u
p
to
1
0
0
0
iter
atio
n
s
[
4
4
]
.
K
-
NN
class
if
ie
s
in
s
tan
ce
s
b
ased
o
n
th
e
m
ajo
r
ity
v
o
te
o
f
th
e
f
iv
e
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J E
lec
&
C
o
m
p
E
n
g
I
SS
N:
2088
-
8
7
0
8
S
tu
d
en
ts
p
erfo
r
ma
n
ce
clu
s
teri
n
g
fo
r
fu
tu
r
e
p
ers
o
n
a
liz
ed
in
…
(
Gh
a
lia
Md
a
g
h
r
i A
la
o
u
i
)
303
n
ea
r
est
n
eig
h
b
o
r
s
,
em
p
lo
y
in
g
u
n
if
o
r
m
weig
h
tin
g
a
n
d
E
u
cl
id
ea
n
d
is
tan
ce
[
4
5
]
.
RF
co
m
b
in
es
1
0
0
d
ec
is
io
n
tr
ee
s
with
m
ajo
r
ity
v
o
tin
g
,
a
m
ax
im
u
m
d
ep
th
o
f
1
0
,
a
n
d
b
o
o
ts
tr
ap
s
am
p
lin
g
to
r
ed
u
ce
o
v
er
f
itti
n
g
an
d
h
ig
h
lig
h
t im
p
o
r
tan
t
f
ea
tu
r
es
[
4
6
]
.
Mo
d
el
tr
ain
in
g
an
d
ev
al
u
atio
n
f
o
llo
wed
a
r
ig
o
r
o
u
s
p
r
o
ce
d
u
r
e
to
en
s
u
r
e
r
eliab
ilit
y
an
d
v
alid
ity
.
T
h
e
d
ataset
was
s
p
lit
in
to
7
0
%
f
o
r
tr
ain
in
g
(
7
0
0
s
tu
d
en
ts
)
an
d
3
0
%
f
o
r
test
in
g
(
3
0
0
s
tu
d
e
n
ts
)
with
s
tr
atif
ied
s
am
p
lin
g
to
m
ain
tain
class
d
i
s
tr
ib
u
tio
n
.
Fiv
e
-
f
o
ld
cr
o
s
s
-
v
al
id
atio
n
was
p
e
r
f
o
r
m
ed
o
n
th
e
tr
ain
in
g
s
et,
an
d
p
er
f
o
r
m
an
ce
was
ass
ess
ed
u
s
i
n
g
ac
cu
r
ac
y
,
F1
-
Sco
r
e
,
an
d
A
UC
-
R
O
C
m
etr
ics.
T
h
is
co
m
p
r
eh
en
s
iv
e
ap
p
r
o
ac
h
en
s
u
r
es
r
o
b
u
s
t
class
if
icatio
n
o
f
s
tu
d
en
t
ca
teg
o
r
ies
wh
ile
lev
er
ag
in
g
t
h
e
clu
s
ter
in
g
-
d
er
iv
e
d
lab
els
to
m
ain
tain
co
n
s
is
ten
cy
with
th
e
d
is
co
v
er
e
d
p
atter
n
s
.
3
.
5
.
I
m
ple
m
ent
a
t
io
n
env
iro
nm
ent
a
nd
t
ec
hn
ica
l sp
ec
if
ic
a
t
io
ns
T
h
e
tech
n
ical
im
p
lem
en
tatio
n
u
s
ed
Py
th
o
n
3
.
1
1
in
a
J
u
p
y
ter
No
teb
o
o
k
en
v
ir
o
n
m
en
t.
T
h
e
m
ain
lib
r
ar
ies
wer
e
Scik
it
-
lear
n
1
.
3
.
0
f
o
r
m
ac
h
in
e
lear
n
in
g
,
Pan
d
a
s
2
.
0
.
3
f
o
r
d
ata
m
an
ip
u
latio
n
,
Nu
m
Py
1
.
2
4
.
3
f
o
r
n
u
m
er
ical
o
p
er
atio
n
s
,
a
n
d
Ma
tp
lo
tlib
3
.
7
.
1
with
Seab
o
r
n
0
.
1
2
.
2
f
o
r
v
is
u
aliza
tio
n
.
A
f
ix
e
d
r
an
d
o
m
s
ee
d
(
4
2
)
en
s
u
r
ed
r
ep
r
o
d
u
ci
b
ilit
y
.
T
h
e
ex
p
er
im
en
tal
m
eth
o
d
o
lo
g
y
f
o
llo
wed
a
s
ix
-
s
tag
e
p
ip
elin
e
in
Fig
u
r
e
2
:
s
tar
tin
g
with
r
aw
C
SV
d
ata,
p
e
r
f
o
r
m
in
g
p
r
ep
r
o
ce
s
s
in
g
an
d
v
ali
d
atio
n
,
a
p
p
ly
in
g
Stan
d
a
r
d
Sc
aler
n
o
r
m
aliza
tio
n
,
ex
ec
u
tin
g
f
iv
e
clu
s
ter
in
g
al
g
o
r
ith
m
s
(
K
-
m
ea
n
s
,
K
-
m
ed
o
id
s
,
GM
M
,
HC
,
an
d
s
p
ec
tr
al
cl
u
s
ter
in
g
)
,
ev
alu
atin
g
with
in
ter
n
al
v
alid
atio
n
m
etr
ic
s
,
an
d
co
n
cl
u
d
in
g
with
v
is
u
aliza
tio
n
an
d
s
tatis
tical
an
aly
s
is
.
T
h
e
co
n
f
ig
u
r
atio
n
p
a
r
am
eter
s
(
Py
th
o
n
3
.
1
1
,
Scik
it
-
lear
n
1
.
3
.
0
,
R
an
d
o
m
Seed
:
4
2
)
ar
e
c
o
n
s
is
ten
tly
ap
p
lied
ac
r
o
s
s
p
r
ep
r
o
ce
s
s
in
g
,
f
ea
tu
r
e
en
g
in
ee
r
in
g
,
a
n
d
clu
s
ter
in
g
s
tag
es
to
en
s
u
r
e
r
ep
r
o
d
u
cib
le
r
esu
lts
an
d
en
ab
le
co
m
p
ar
ativ
e
an
al
y
s
is
b
etwe
en
d
if
f
er
en
t a
l
g
o
r
ith
m
ic
a
p
p
r
o
ac
h
es.
Fig
u
r
e
2
.
E
x
p
er
im
e
n
tal
p
ip
elin
e
f
r
o
m
p
r
ep
r
o
ce
s
s
in
g
to
ev
al
u
atio
n
an
d
v
is
u
aliza
tio
n
in
Py
th
o
n
3
.
1
1
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8
7
0
8
I
n
t J E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
1
6
,
No
.
1
,
Feb
r
u
ar
y
20
2
6
:
2
9
7
-
310
304
4.
RE
SU
L
T
S AN
D
D
I
SCU
SS
I
O
N
4
.
1
.
Clus
t
er
ing
re
s
ults o
v
er
v
iew
T
h
e
ad
v
an
ce
d
clu
s
ter
in
g
&
d
a
ta
v
is
u
aliza
tio
n
s
u
ite
is
a
tech
n
ical
f
r
am
ewo
r
k
th
at
allo
ws
r
esear
ch
er
s
to
im
p
o
r
t
C
SV
d
atasets
an
d
ap
p
ly
f
iv
e
clu
s
ter
in
g
alg
o
r
ith
m
s
(
K
-
Me
an
s
,
K
-
Me
d
o
id
s
,
GM
M
,
HC
,
an
d
s
p
ec
tr
al
clu
s
ter
in
g
)
to
an
aly
ze
ac
ad
em
ic
p
er
f
o
r
m
a
n
ce
d
ata.
Ad
v
a
n
ce
d
m
eth
o
d
s
,
s
u
ch
as
HPEFC
M
-
FS
P
f
o
r
clu
s
ter
in
g
an
d
Neu
r
o
E
v
o
C
lass
f
o
r
p
r
ed
ictiv
e
m
o
d
elin
g
,
c
an
b
e
e
m
p
lo
y
ed
t
o
id
e
n
tify
h
ig
h
-
ac
h
ie
v
in
g
,
av
er
a
g
e,
a
n
d
s
tr
u
g
g
lin
g
s
tu
d
en
ts
,
en
a
b
lin
g
d
ata
-
d
r
iv
e
n
in
ter
v
en
tio
n
s
.
Op
tim
ized
v
ia
p
ar
ticle
s
war
m
o
p
tim
izatio
n
(
PSO
)
an
d
ar
tific
ial
n
eu
r
al
n
etwo
r
k
(
ANN
)
,
th
ese
alg
o
r
ith
m
s
en
h
an
ce
ac
cu
r
ac
y
,
p
r
ec
is
io
n
,
an
d
r
ec
a
ll,
s
u
p
p
o
r
tin
g
ea
r
ly
war
n
in
g
s
y
s
tem
s
an
d
p
er
s
o
n
al
ized
lear
n
in
g
p
ath
wa
y
s
,
s
im
il
ar
to
th
e
p
r
ed
ictiv
e
ap
p
r
o
ac
h
s
u
g
g
ested
b
y
Ma
lik
et
a
l
[
4
7
]
.
T
h
e
m
o
d
u
lar
in
ter
f
ac
e
in
clu
d
es
a
lef
t
p
an
el
f
o
r
d
ata
m
an
ag
em
en
t
an
d
C
SV
im
p
o
r
t,
alg
o
r
ith
m
-
s
p
ec
if
ic
tab
s
f
o
r
a
n
aly
s
is
,
an
d
in
teg
r
ated
r
esu
lts
co
m
p
a
r
is
o
n
u
s
in
g
Da
v
ies
-
B
o
u
ld
in
a
n
d
C
alin
s
k
i
-
Har
ab
asz
p
er
f
o
r
m
an
ce
m
etr
ics,
cr
ea
tin
g
a
u
n
if
ied
an
aly
tical
p
ip
elin
e
f
o
r
ed
u
ca
tio
n
al
d
ata
m
in
in
g
.
Fig
u
r
e
3
illu
s
tr
ates
th
e
co
m
p
lete
in
ter
f
ac
e
ar
ch
ite
ctu
r
e
an
d
w
o
r
k
f
l
o
w
im
p
lem
en
tatio
n
.
Fig
u
r
e
3
.
Mu
lti
-
alg
o
r
ith
m
cl
u
s
ter
in
g
an
aly
s
is
p
latf
o
r
m
in
ter
f
ac
e
4
.
2
.
K
-
M
ea
ns
clus
t
er
ing
per
f
o
rm
a
nce
a
na
ly
s
is
T
h
e
ap
p
licatio
n
o
f
th
e
K
-
M
ea
n
s
alg
o
r
ith
m
r
ev
ea
led
cr
it
ical
in
s
ig
h
ts
in
to
s
tu
d
en
t
p
er
f
o
r
m
a
n
ce
p
atter
n
s
,
d
em
o
n
s
tr
atin
g
o
p
tim
al
clu
s
ter
in
g
p
er
f
o
r
m
an
ce
at
k
=
3
clu
s
ter
s
.
T
h
is
th
r
ee
-
clu
s
ter
s
o
lu
tio
n
r
ef
lects
co
m
m
o
n
e
d
u
ca
tio
n
al
p
r
ac
tices
o
f
g
r
o
u
p
in
g
lear
n
e
r
s
in
to
b
eg
in
n
er
,
in
ter
m
ed
iate,
an
d
a
d
v
an
ce
d
lev
els.
T
h
e
alg
o
r
ith
m
'
s
ev
alu
atio
n
m
etr
ics,
in
clu
d
i
n
g
a
Dav
ies
-
B
o
u
ld
in
i
n
d
ex
(
DB
I
)
o
f
0
.
7
9
2
3
a
n
d
a
C
alin
s
k
i
-
Har
ab
asz
i
n
d
ex
(
C
HI
)
o
f
1
3
9
8
,
in
d
icate
s
u
b
s
tan
tial
im
p
r
o
v
em
en
ts
in
clu
s
ter
in
g
q
u
ality
co
m
p
a
r
ed
t
o
p
r
ev
io
u
s
s
tu
d
ies.
No
tab
ly
,
o
u
r
k
=
3
s
o
lu
tio
n
ef
f
ec
tiv
ely
ad
d
r
ess
es
th
e
g
r
an
u
lar
ity
v
er
s
u
s
p
r
ac
ticality
tr
ad
e
-
o
f
f
th
at
h
as
ch
allen
g
ed
ed
u
ca
tio
n
al
clu
s
ter
in
g
ap
p
licatio
n
s
,
p
r
o
v
i
d
in
g
s
u
f
f
icien
t
d
etail
f
o
r
p
er
s
o
n
alize
d
in
ter
v
en
tio
n
s
wh
ile
m
ain
tain
in
g
m
an
ag
ea
b
le
im
p
lem
en
tatio
n
co
m
p
le
x
ity
.
T
h
e
r
esu
ltin
g
clu
s
ter
s
ar
e
v
is
u
ally
r
ep
r
esen
ted
in
Fig
u
r
e
4
,
wh
ile
th
e
co
r
r
esp
o
n
d
in
g
ev
alu
atio
n
m
etr
ics
ar
e
s
u
m
m
ar
ized
in
th
e
d
e
d
icate
d
r
es
u
lts
co
m
p
ar
is
o
n
tab
o
f
th
e
in
ter
f
ac
e.
Similar
an
aly
tical
p
r
o
ce
d
u
r
es
wer
e
ap
p
lied
to
all
clu
s
ter
in
g
alg
o
r
ith
m
s
to
en
ab
le
a
co
m
p
r
eh
e
n
s
iv
e
p
er
f
o
r
m
a
n
ce
c
o
m
p
ar
is
o
n
.
4
.
3
.
Co
m
pa
ra
t
iv
e
a
na
ly
s
is
a
nd
cr
it
ica
l int
er
pret
a
t
io
n o
f
clus
t
er
ing
re
s
u
lt
s
W
e
ev
alu
ated
f
iv
e
clu
s
ter
in
g
alg
o
r
ith
m
s
—
s
p
ec
tr
al
clu
s
ter
in
g
,
K
-
Me
a
n
s
,
K
-
Me
d
o
id
s
,
Gau
s
s
ian
m
ix
tu
r
e
m
o
d
el
(
GM
M)
,
an
d
Hier
ar
ch
ical
C
lu
s
ter
in
g
—
an
d
f
o
u
n
d
th
at
Sp
ec
tr
al
C
lu
s
ter
in
g
p
er
f
o
r
m
e
d
b
est
(
Dav
ies
-
B
o
u
ld
in
I
n
d
ex
:
0
.
7
5
6
9
,
C
alin
s
k
i
-
Har
ab
asz
I
n
d
ex
:
1
3
2
2
.
4
2
2
)
,
f
o
llo
wed
clo
s
ely
b
y
K
-
Me
a
n
s
(
DB
I
:
0
.
7
9
2
3
,
C
HI
:
1
3
9
8
.
4
6
2
3
)
.
B
o
th
d
em
o
n
s
tr
ated
s
tr
o
n
g
clu
s
te
r
in
g
q
u
ality
,
as
DB
I
v
alu
es
b
elo
w
1
.
0
in
d
icate
co
m
p
ac
t
an
d
well
-
s
ep
ar
ated
cl
u
s
ter
s
.
C
o
m
p
ar
ed
to
[
4
1
]
,
wh
i
ch
r
ep
o
r
ted
a
K
-
Me
an
s
DB
I
o
f
1
.
7
1
,
o
u
r
s
p
ec
tr
al
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J E
lec
&
C
o
m
p
E
n
g
I
SS
N:
2088
-
8
7
0
8
S
tu
d
en
ts
p
erfo
r
ma
n
ce
clu
s
teri
n
g
fo
r
fu
tu
r
e
p
ers
o
n
a
liz
ed
in
…
(
Gh
a
lia
Md
a
g
h
r
i A
la
o
u
i
)
305
ap
p
r
o
ac
h
s
h
o
ws
a
5
6
%
im
p
r
o
v
em
en
t,
co
n
f
ir
m
in
g
its
ef
f
ec
tiv
en
ess
in
ca
p
tu
r
i
n
g
n
o
n
-
lin
ea
r
p
atter
n
s
i
n
ed
u
ca
tio
n
al
d
ata.
K
-
Me
d
o
id
s
y
ield
ed
s
lig
h
tly
h
ig
h
er
v
alu
es
(
DB
I
:
0
.
8
1
1
5
,
C
HI
:
1
3
6
3
.
3
1
9
5
)
,
wh
ile
GM
M
ac
h
iev
ed
co
m
p
ar
ab
le
p
er
f
o
r
m
an
c
e
(
DB
I
:
0
.
8
0
1
1
,
C
HI
:
1
3
5
0
.
3
1
4
8
)
.
Hier
ar
ch
ical
C
lu
s
ter
in
g
p
r
o
d
u
ce
d
th
e
h
i
g
h
est
DB
I
(
0
.
8
2
9
7
)
a
n
d
th
e
lo
west
C
HI
(
1
1
8
9
.
2
6
5
7
)
,
s
u
g
g
esti
n
g
wea
k
er
s
ep
ar
atio
n
b
etwe
en
clu
s
ter
s
.
Ov
er
all,
s
in
ce
a
lo
wer
DB
I
an
d
h
ig
h
er
C
HI
in
d
icate
b
etter
-
d
ef
i
n
ed
clu
s
ter
s
,
s
p
ec
tr
al
clu
s
ter
in
g
em
er
g
ed
as
th
e
m
o
s
t
ef
f
e
ctiv
e
alg
o
r
ith
m
f
o
r
th
is
d
ataset.
Fig
u
r
e
5
p
r
esen
ts
th
e
ev
alu
atio
n
r
esu
lts
f
o
r
all
clu
s
ter
in
g
alg
o
r
it
h
m
s
.
Fig
u
r
e
4
.
K
-
Me
a
n
s
clu
s
ter
in
g
v
is
u
aliza
tio
n
o
f
s
tu
d
en
t
p
er
f
o
r
m
an
ce
d
ata
Fig
u
r
e
5
.
C
lu
s
ter
in
g
e
v
alu
atio
n
r
esu
lts
f
o
r
th
e
f
iv
e
test
ed
alg
o
r
ith
m
s
T
h
e
s
u
p
e
r
io
r
p
er
f
o
r
m
an
ce
o
f
s
p
ec
tr
al
cl
u
s
ter
in
g
r
ep
r
ese
n
ts
a
p
a
r
ad
ig
m
s
h
if
t
f
r
o
m
tr
ad
itio
n
al
d
is
tan
ce
-
b
ased
clu
s
ter
in
g
ap
p
r
o
ac
h
es
in
ed
u
ca
tio
n
al
d
ata
m
in
in
g
.
W
h
ile
L
iu
et
a
l
.
[
4
8
]
d
em
o
n
s
tr
ated
th
e
ef
f
ec
tiv
en
ess
o
f
K
-
Me
an
s
u
n
d
er
o
p
tim
al
c
o
n
d
itio
n
s
;
r
ec
en
t
s
tu
d
ies
s
u
g
g
est
th
at
s
p
ec
tr
a
l
m
eth
o
d
s
u
n
c
o
v
er
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8
7
0
8
I
n
t J E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
1
6
,
No
.
1
,
Feb
r
u
ar
y
20
2
6
:
2
9
7
-
310
306
d
ee
p
er
s
tr
u
ctu
r
e
i
n
s
tu
d
en
t
d
a
ta.
Fo
r
in
s
tan
ce
,
Qu
y
et
a
l
.
[
4
9
]
r
ev
iew
th
e
ed
u
ca
tio
n
al
d
at
a
s
cien
ce
f
ield
an
d
h
ig
h
lig
h
t
t
h
e
n
ee
d
f
o
r
ad
v
an
c
ed
clu
s
ter
in
g
m
eth
o
d
s
to
en
s
u
r
e
b
o
th
ac
cu
r
ac
y
an
d
f
air
n
ess
in
s
tu
d
en
t
p
r
o
f
ilin
g
,
an
d
Yu
et
a
l
.
[
3
6
]
d
e
m
o
n
s
tr
at
e
th
at
ad
ap
tiv
e
f
u
zz
y
s
p
ec
tr
al
clu
s
ter
in
g
s
ig
n
if
ican
tly
en
h
a
n
ce
s
clu
s
ter
q
u
ality
o
n
c
o
m
p
lex
n
o
n
lin
ea
r
d
atasets
.
Sp
ec
tr
al
c
lu
s
ter
in
g
,
with
it
s
g
r
ap
h
-
b
ased
s
im
ilar
ity
ap
p
r
o
ac
h
,
t
h
u
s
en
a
b
les
id
en
tific
atio
n
o
f
s
u
b
tle
p
er
f
o
r
m
an
ce
r
elatio
n
s
h
ip
s
th
at
ce
n
tr
o
id
-
b
ased
m
eth
o
d
s
lik
e
K
-
Me
an
s
m
ay
m
is
s
.
T
h
i
s
f
in
d
in
g
c
h
allen
g
es
th
e
p
r
ev
ale
n
t
u
s
e
o
f
K
-
Me
an
s
in
ed
u
ca
ti
o
n
al
clu
s
ter
in
g
a
n
d
s
u
g
g
ests
th
at
th
e
ed
u
ca
tio
n
al
r
esear
ch
co
m
m
u
n
ity
s
h
o
u
ld
ad
o
p
t sp
ec
tr
al
m
eth
o
d
s
f
o
r
m
o
r
e
ac
cu
r
ate
s
tu
d
en
t p
r
o
f
ilin
g
.
T
h
e
4
% im
p
r
o
v
e
m
en
t
in
DB
I
s
co
r
es
(
0
.
7
5
v
s
0
.
7
9
)
m
ay
ap
p
ea
r
m
o
d
est,
b
u
t
it
r
ep
r
esen
ts
s
u
b
s
tan
tial
p
r
ac
tical
s
ig
n
if
ican
ce
wh
en
ap
p
lied
to
lar
g
e
-
s
ca
le
ed
u
c
atio
n
al
s
y
s
tem
s
wh
er
e
im
p
r
o
v
ed
cl
u
s
ter
in
g
ac
cu
r
ac
y
d
ir
ec
tly
in
f
lu
e
n
ce
s
p
er
s
o
n
aliza
tio
n
ef
f
ec
tiv
en
ess
.
4
.
3
.
Student
clus
t
er
a
s
s
ig
nm
ent
a
nd
perf
o
rm
a
nce
la
belin
g
T
h
e
clu
s
ter
in
g
r
esu
lts
r
ev
e
al
th
at
th
r
ee
d
is
tin
ct
s
tu
d
en
t
g
r
o
u
p
s
em
er
g
ed
n
atu
r
all
y
-
with
o
u
t
p
r
esu
p
p
o
s
in
g
th
eir
n
u
m
b
er
-
co
n
s
is
ten
t
with
ex
p
lo
r
ato
r
y
ed
u
c
atio
n
al
r
esear
ch
in
d
icatin
g
s
i
m
ilar
s
tr
u
ctu
r
es.
Fo
r
in
s
tan
ce
,
W
o
o
d
s
et
a
l.
[
5
0
]
a
p
p
lied
a
cl
u
s
ter
an
aly
s
is
to
ea
r
ly
elem
en
tar
y
s
tu
d
en
t
d
ata
a
n
d
f
o
u
n
d
th
at
th
r
ee
clu
s
ter
s
(
lo
w,
av
er
a
g
e,
a
n
d
h
i
g
h
p
e
r
f
o
r
m
er
s
)
r
ep
r
esen
ted
m
ea
n
in
g
f
u
l
lear
n
in
g
s
u
b
g
r
o
u
p
s
,
r
ath
er
th
an
r
ely
in
g
o
n
ar
b
itra
r
y
class
if
icatio
n
s
.
T
h
is
ap
p
r
o
ac
h
alig
n
s
with
m
et
h
o
d
o
lo
g
ical
r
ec
o
m
m
en
d
atio
n
s
in
th
e
ed
u
ca
tio
n
al
d
ata
m
in
in
g
liter
atu
r
e,
wh
er
e
s
elec
tin
g
th
r
ee
clu
s
ter
s
o
f
ten
b
a
lan
ce
s
in
ter
p
r
etab
ilit
y
an
d
s
tatis
tical
v
alid
ity
.
C
o
n
s
eq
u
en
tly
,
lab
elin
g
th
e
r
e
s
u
ltin
g
g
r
o
u
p
s
as
ad
v
a
n
ce
d
,
i
n
ter
m
ed
iate,
an
d
b
e
g
in
n
er
is
s
u
p
p
o
r
ted
b
o
th
b
y
o
u
r
em
p
i
r
ical
f
in
d
in
g
s
an
d
b
y
p
r
io
r
s
tu
d
ies
s
u
g
g
esti
n
g
th
at
s
tu
d
en
t
p
er
f
o
r
m
an
ce
n
atu
r
ally
o
r
g
an
izes
in
to
th
r
ee
lev
els.
T
h
is
class
if
icatio
n
s
y
s
tem
r
ef
lects
a
ctu
al
co
m
p
eten
cy
tier
s
m
o
r
e
ac
cu
r
ately
th
an
co
n
v
en
tio
n
al
g
r
ad
in
g
r
e
g
im
es.
As
illu
s
tr
ated
in
Fig
u
r
e
6
,
t
h
e
th
r
ee
-
clu
s
ter
s
tr
u
ctu
r
e
p
r
o
v
id
es
a
clea
r
v
is
u
al
s
ep
ar
atio
n
o
f
s
tu
d
en
t
g
r
o
u
p
s
,
r
ein
f
o
r
cin
g
th
e
v
alid
ity
o
f
th
is
ca
teg
o
r
izatio
n
b
ased
o
n
le
ar
n
in
g
p
atter
n
s
an
d
p
er
f
o
r
m
an
ce
d
ata.
Fig
u
r
e
6
.
Stu
d
e
n
t
clu
s
ter
ass
ig
n
m
en
ts
an
d
p
er
f
o
r
m
an
ce
class
if
icatio
n
r
esu
lts
4
.
4
.
Cla
s
s
if
ica
t
io
n
m
o
del per
f
o
rma
nce
a
nd
predict
iv
e
a
cc
ura
cy
Ou
r
class
if
icatio
n
ev
alu
atio
n
u
s
in
g
f
iv
e
m
ac
h
in
e
-
lear
n
in
g
m
o
d
els
d
em
o
n
s
tr
ates
ex
ce
p
tio
n
al
p
r
ed
ictiv
e
ca
p
a
b
ilit
y
th
at
s
u
b
s
tan
tially
s
u
r
p
ass
es
p
r
ev
io
u
s
e
d
u
ca
tio
n
al
class
if
icatio
n
s
tu
d
i
es.
T
ab
le
3
d
is
p
lay
s
th
e
p
er
f
o
r
m
an
ce
m
etr
ics
o
b
tai
n
ed
ac
r
o
s
s
th
e
f
iv
e
m
o
d
els,
h
i
g
h
lig
h
tin
g
t
h
eir
s
tr
o
n
g
a
b
ilit
y
to
p
r
ed
ict
clu
s
ter
-
b
ased
p
er
f
o
r
m
an
ce
lab
els with
h
ig
h
ac
c
u
r
ac
y
a
n
d
r
eliab
ilit
y
.
T
ab
le
3
.
Per
f
o
r
m
an
ce
o
f
class
if
icatio
n
m
o
d
els in
p
r
ed
ictin
g
s
tu
d
en
t c
lu
s
ter
lab
els
M
o
d
e
l
A
c
c
u
r
a
c
y
F1
-
S
c
o
r
e
AUC
-
R
O
C
Lo
g
i
s
t
i
c
r
e
g
r
e
ssi
o
n
0
.
9
9
0
0
.
9
9
0
0
2
2
0
.
9
9
9
7
5
1
S
V
M
0
.
9
7
5
0
.
9
7
5
0
2
0
0
.
9
9
9
7
8
7
K
N
N
0
.
9
7
0
0
.
9
6
9
8
7
8
0
.
9
9
8
7
8
7
R
a
n
d
o
m f
o
r
e
s
t
0
.
9
4
5
0
.
9
4
5
0
2
5
0
.
9
9
6
9
9
5
D
e
c
i
s
i
o
n
t
r
e
e
0
.
9
4
5
0
.
9
4
4
9
0
9
0
.
9
5
8
7
9
5
T
h
e
r
esu
lts
o
f
o
u
r
s
tu
d
y
d
em
o
n
s
tr
ate
th
at
class
if
icatio
n
m
o
d
els
ap
p
lied
to
th
e
clu
s
ter
lab
el
s
ac
h
iev
e
h
ig
h
p
e
r
f
o
r
m
an
ce
,
with
ac
cu
r
ac
ies
r
an
g
in
g
f
r
o
m
9
4
.
5
%
f
o
r
tr
ee
-
b
ased
m
o
d
els
(
R
an
d
o
m
Fo
r
est,
Dec
is
io
n
T
r
ee
)
u
p
to
9
9
%
f
o
r
LR
,
alo
n
g
s
id
e
ex
ce
llen
t
F1
-
S
co
r
es
a
n
d
AUC
-
R
O
C
v
alu
es.
T
h
ese
o
u
t
co
m
es
co
n
f
ir
m
th
e
q
u
ality
o
f
th
e
p
r
io
r
u
n
s
u
p
er
v
is
ed
clu
s
ter
in
g
s
tep
an
d
th
e
ab
ili
ty
o
f
s
u
p
er
v
is
ed
al
g
o
r
ith
m
s
to
ef
f
ec
tiv
ely
p
r
ed
ict
s
tu
d
en
t g
r
o
u
p
s
.
Evaluation Warning : The document was created with Spire.PDF for Python.