Co
m
pu
t
er
Science
a
nd
I
nfo
r
m
a
t
io
n T
ec
hn
o
lo
g
ies
Vo
l.
7
,
No
.
2
,
J
u
ly
20
26
,
p
p
.
1
7
9
~
1
95
I
SS
N:
2722
-
3
2
2
1
,
DOI
:
1
0
.
1
1
5
9
1
/csi
t
.
v
7
i
2
.
p
1
7
9
-
1
95
179
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ia
e
s
p
r
ime.
co
m/in
d
ex
.
p
h
p
/csi
t
A comp
a
ra
tive st
udy
of clas
sica
l,
b
a
g
g
ing
,
a
nd hy
brid methods
for
o
ptimizing
loa
n def
a
ul
t
pre
dicti
o
n
I
s
m
a
il Id
o
wu A
k
uji
1
,
Ahm
ed
B
a
ba
j
ide O
la
nrewa
j
u
2
,
T
a
o
f
ik
Abio
du
n Ahm
ed
3
,
Ay
o
dej
i J
ub
ril
Ala
bi
4
,
I
dris
B
a
ba
t
un
de
A
dey
em
i
5
1
D
e
p
a
r
t
me
n
t
o
f
C
o
m
p
u
t
e
r
S
c
i
e
n
c
e
,
F
a
c
u
l
t
y
o
f
S
c
i
e
n
c
e
a
n
d
C
o
m
p
u
t
i
n
g
,
A
b
d
u
l
r
a
s
a
q
A
b
u
b
a
k
a
r
To
y
i
n
U
n
i
v
e
r
s
i
t
y
,
O
k
e
-
O
g
b
a
,
N
i
g
e
r
i
a
2
D
e
p
a
r
t
me
n
t
o
f
C
o
m
p
u
t
e
r
S
c
i
e
n
c
e
,
U
n
i
v
e
r
s
i
t
y
o
f
I
b
a
d
a
n
,
I
b
a
d
a
n
,
N
i
g
e
r
i
a
3
D
e
p
a
r
t
me
n
t
o
f
C
o
m
p
u
t
e
r
S
c
i
e
n
c
e
,
S
c
h
o
o
l
o
f
I
C
T
a
n
d
M
a
n
a
g
e
me
n
t
S
c
i
e
n
c
e
,
K
w
a
r
a
S
t
a
t
e
C
o
l
l
e
g
e
o
f
A
r
a
b
i
c
a
n
d
I
sl
a
m
i
c
Le
g
a
l
S
t
u
d
i
e
s,
I
l
o
r
i
n
,
N
i
g
e
r
i
a
4
D
e
p
a
r
t
me
n
t
o
f
C
o
m
p
u
t
e
r
S
c
i
e
n
c
e
,
F
a
c
u
l
t
y
o
f
I
n
f
o
r
ma
t
i
o
n
a
n
d
C
o
mm
u
n
i
c
a
t
i
o
n
T
e
c
h
n
o
l
o
g
y
,
K
w
a
r
a
S
t
a
t
e
U
n
i
v
e
r
s
i
t
y
,
M
a
l
e
t
e
,
N
i
g
e
r
i
a
5
D
e
p
a
r
t
me
n
t
o
f
C
o
m
p
u
t
e
r
S
c
i
e
n
c
e
s,
F
a
c
u
l
t
y
o
f
C
o
mm
u
n
i
c
a
t
i
o
n
a
n
d
I
n
f
o
r
m
a
t
i
o
n
S
c
i
e
n
c
e
s
,
U
n
i
v
e
r
s
i
t
i
o
f
I
l
o
r
i
n
,
I
l
o
r
i
n
,
N
i
g
e
r
i
a
Art
icle
I
nfo
AB
S
T
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
Oct
2
4
,
2
0
2
5
R
ev
is
ed
Ma
r
1
1
,
2
0
2
6
Acc
ep
ted
Ma
y
1
6
,
2
0
2
6
Th
is
stu
d
y
o
p
ti
m
ize
d
lo
a
n
d
e
fa
u
lt
p
re
d
icti
o
n
b
y
c
o
m
p
a
ri
n
g
k
-
n
e
a
re
st
n
e
ig
h
b
o
r
(KN
N),
ra
n
d
o
m
fo
re
st
(
RF
),
a
n
d
h
y
b
ri
d
m
e
th
o
d
s.
Th
e
d
a
tas
e
t
u
se
d
wa
s
p
re
p
ro
c
e
ss
e
d
u
sin
g
sim
p
le
i
m
p
u
ter,
lab
e
l
e
n
c
o
d
e
r,
s
y
n
t
h
e
ti
c
m
in
o
rit
y
o
v
e
rsa
m
p
li
n
g
tec
h
n
iq
u
e
(S
M
OT
E),
a
n
d
c
o
rre
l
a
ti
o
n
-
b
a
se
d
fe
a
tu
re
se
lec
ti
o
n
o
n
to
p
7
fe
a
tu
re
s
wh
il
e
g
r
id
se
a
rc
h
c
ro
ss
-
v
a
li
d
a
ti
o
n
(G
S
CV)
a
n
d
ra
n
d
o
m
se
a
rc
h
c
ro
ss
-
v
a
li
d
a
ti
o
n
(RS
CV)
we
re
e
m
p
lo
y
e
d
t
o
o
p
ti
m
ize
m
o
d
e
ls.
Be
fo
re
tu
n
i
n
g
,
RF
a
c
h
iev
e
d
p
e
rfe
c
t
p
e
rf
o
rm
a
n
c
e
(1
0
0
%
a
c
c
u
ra
c
y
,
9
9
.
8
%
p
re
c
isio
n
,
1
0
0
%
re
c
a
l
l,
9
9
.
9
%
F
1
,
1
.
0
0
0
a
re
a
u
n
d
e
r
c
u
r
v
e
(
AUC
)
),
o
u
tp
e
rfo
rm
in
g
u
n
t
u
n
e
d
KN
N
(9
9
.
2
%
a
c
c
u
ra
c
y
,
9
6
.
2
%
p
re
c
isio
n
,
9
9
.
8
%
re
c
a
ll
,
9
8
.
0
%
F
1
,
0
.
9
9
7
AU
C)
a
n
d
h
y
b
r
id
(9
9
.
8
%
a
c
c
u
ra
c
y
,
9
9
.
1
%
p
re
c
isio
n
,
9
9
.
9
%
re
c
a
ll
,
9
9
.
5
%
F
1
)
.
Afte
r
t
u
n
i
n
g
,
RF
m
a
in
tain
e
d
sa
m
e
re
su
lt
s,
c
o
n
fi
rm
e
d
b
y
1
0
×
n
e
ste
d
CV
sta
b
il
it
y
(F
1
=
0
.
9
9
9
7
±0
.
0
0
0
2
)
a
n
d
M
c
Ne
m
a
r
tes
ts
sh
o
wi
n
g
e
q
u
iv
a
len
c
e
to
RF
_
RS
CV
(
p
=
1
.
0
0
0
0
).
KN
N
imp
r
o
v
e
d
m
a
rg
in
a
ll
y
i
n
p
re
c
isio
n
(
9
6
.
2
%
→
9
9
.
8
%
)
b
u
t
d
e
c
li
n
e
d
i
n
re
c
a
ll
,
wh
il
e
h
y
b
r
i
d
d
r
o
p
p
e
d
slig
h
tl
y
a
c
ro
ss
m
e
tri
c
s.
P
a
rti
a
l
d
e
p
e
n
d
e
n
c
e
p
l
o
t
s
c
o
n
firm
RF
’
s
d
o
m
in
a
n
c
e
ste
m
s
fro
m
th
re
e
k
e
y
fe
a
tu
re
s
(lu
m
p
_
s
u
m
_
p
a
y
m
e
n
t,
p
r
o
p
e
rt
y
_
v
a
lu
e
,
c
o
-
a
p
p
li
c
a
n
t
_
c
re
d
it
_
ty
p
e
),
v
a
li
d
a
ted
b
y
b
u
sin
e
ss
imp
a
c
t
a
n
a
ly
sis
sh
o
win
g
m
in
ima
l
e
rro
rs
a
g
a
in
st
KN
N/
h
y
b
rid
.
RF
_
G
S
CV
’
s
p
e
rfe
c
ti
o
n
re
f
lec
ts
tru
e
g
e
n
e
ra
li
z
a
ti
o
n
,
n
o
t
o
v
e
rfit
ti
n
g
,
e
sta
b
li
sh
in
g
i
t
a
s
th
e
p
ro
d
u
c
ti
o
n
-
r
e
a
d
y
g
o
ld
sta
n
d
a
rd
.
F
u
tu
re
wo
r
k
c
a
n
a
d
d
re
ss
sta
ti
c
d
a
tas
e
t
l
imitatio
n
b
y
i
n
c
o
rp
o
ra
ti
n
g
d
y
n
a
m
ic
ti
m
e
-
se
ries
d
a
ta
with
o
n
li
n
e
lea
rn
in
g
,
c
o
n
c
e
p
t
d
rift
d
e
te
c
ti
o
n
,
a
n
d
re
a
l
-
ti
m
e
m
a
c
ro
e
c
o
n
o
m
ic fea
tu
re
s to
e
n
h
a
n
c
e
re
a
l
-
wo
rld
g
e
n
e
ra
li
z
a
b
il
it
y
.
K
ey
w
o
r
d
s
:
C
o
r
r
elatio
n
b
ased
f
ea
tu
r
e
s
elec
tio
n
Gr
id
s
ea
r
ch
cr
o
s
s
-
v
alid
atio
n
K
-
n
ea
r
est n
eig
h
b
o
r
L
o
an
d
e
f
au
lt
R
an
d
o
m
f
o
r
est
R
an
d
o
m
s
ea
r
ch
cr
o
s
s
-
v
alid
atio
n
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
s
m
ail
I
d
o
wu
Ak
u
ji
Dep
ar
tm
en
t o
f
C
o
m
p
u
ter
Scie
n
ce
,
Facu
lty
o
f
Scien
ce
an
d
C
o
m
p
u
tin
g
Ab
d
u
lr
asaq
Ab
u
b
ak
ar
T
o
y
i
n
Un
iv
er
s
ity
Ok
e
-
Og
b
a,
Kwa
r
a
State,
Nig
e
r
ia
E
m
ail: a
k
u
jiis
m
ah
ee
l@
g
m
ail.
co
m
1.
I
NT
RO
D
UCT
I
O
N
L
en
d
in
g
an
d
b
o
r
r
o
win
g
am
o
n
g
p
eo
p
le
ar
e
p
r
ev
alen
t
d
u
e
to
f
in
an
cial
in
s
ec
u
r
ity
an
d
ec
o
n
o
m
ic
h
ar
d
s
h
ip
,
p
ar
ticu
lar
l
y
in
Nig
er
ia
wh
er
e
p
o
v
e
r
ty
c
y
cles
p
er
s
is
t
[
1
]
.
Acc
o
r
d
in
g
to
[
2
]
,
lo
an
r
eq
u
est
h
as
tr
em
en
d
o
u
s
ly
in
cr
ea
s
ed
s
in
ce
th
e
C
OVI
D
-
1
9
p
an
d
em
ic,
m
ak
in
g
t
h
e
n
u
m
b
er
o
f
licen
s
ed
lo
an
co
m
p
a
n
ies
r
is
in
g
to
2
8
4
as
o
f
Ma
y
2
0
2
4
[
3
]
.
T
h
is
s
u
r
g
e
in
d
em
an
d
h
as
led
to
a
s
ig
n
if
ican
t
in
c
r
ea
s
e
in
lo
an
d
ef
au
lts
,
r
esu
ltin
g
in
s
u
b
s
tan
tial f
in
an
ci
al
lo
s
s
es f
o
r
f
in
an
cial
in
s
titu
tio
n
s
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
7
2
2
-
3
2
2
1
C
o
m
p
u
t Sci
I
n
f
T
ec
h
n
o
l
,
Vo
l.
7
,
No
.
2
,
J
u
ly
20
26
:
179
-
1
95
180
L
o
an
d
e
f
au
lt
h
as
b
ec
o
m
e
a
d
o
m
in
an
t
is
s
u
e,
r
esu
ltin
g
f
r
o
m
i
n
ad
eq
u
ate
l
o
an
m
a
n
ag
em
en
t
s
y
s
tem
s
[
4
]
,
h
ig
h
lig
h
tin
g
th
e
n
ee
d
f
o
r
f
i
n
an
cial
in
s
titu
tio
n
s
to
im
p
r
o
v
e
lo
an
m
a
n
ag
em
en
t.
L
o
an
d
ef
au
lt
p
r
ed
ictio
n
h
as
b
ec
o
m
e
a
cr
itical
co
n
ce
r
n
f
o
r
f
in
an
cial
in
s
titu
tio
n
s
an
d
len
d
er
s
wo
r
ld
wid
e,
as
d
ef
au
ltin
g
o
n
a
lo
a
n
ca
n
lead
to
f
in
an
cial
d
is
tr
ess
,
r
ep
u
tatio
n
al
d
am
ag
e,
an
d
o
p
er
atio
n
al
d
if
f
icu
lties
[
5
]
.
Mo
s
t
o
f
th
e
co
n
v
en
tio
n
al
lo
a
n
p
r
ed
ictio
n
m
o
d
els
p
r
ed
o
m
in
a
n
tly
em
p
lo
y
tr
a
d
itio
n
al
alg
o
r
it
h
m
s
in
clu
d
in
g
r
a
n
d
o
m
f
o
r
est
(
R
F),
d
ec
is
io
n
tr
ee
s
(
DT
)
,
n
aïv
e
B
ay
es (
NB
)
an
d
s
u
p
p
o
r
t v
ec
to
r
m
ac
h
in
es (
SVM
s
)
[
6
]
−
[
8
]
.
Ad
d
itio
n
ally
,
k
-
n
ea
r
est
n
eig
h
b
o
r
s
(
KNN)
h
av
e
d
em
o
n
s
tr
ated
s
tr
o
n
g
p
r
ed
ictiv
e
p
o
w
er
in
l
o
an
ap
p
r
o
v
al
task
s
,
with
th
e
p
o
ten
tial
to
r
ev
o
lu
tio
n
ize
lo
an
ap
p
r
o
v
al
p
r
o
ce
s
s
es
in
f
in
a
n
cial
in
s
titu
tio
n
s
.
A
r
ec
en
t
s
tu
d
y
b
y
[
9
]
f
o
u
n
d
th
at
KNN
o
u
tp
er
f
o
r
m
ed
o
th
e
r
alg
o
r
ith
m
s
,
in
clu
d
in
g
RF
,
SVM,
lin
ea
r
r
eg
r
ess
io
n
(
L
R
)
,
a
n
d
d
ec
is
io
n
tr
ee
(
DT
)
,
ac
h
iev
in
g
o
p
tim
al
ac
cu
r
ac
y
.
C
o
n
s
id
e
r
in
g
th
ese
r
esu
lts
,
a
h
y
b
r
id
a
p
p
r
o
ac
h
co
m
b
in
in
g
KNN
with
R
F
ca
n
en
h
an
ce
th
e
r
o
b
u
s
tn
ess
o
f
lo
an
p
r
ed
ictiv
e
m
o
d
els
th
r
o
u
g
h
KNN
’
s
ab
ilit
y
to
ca
p
tu
r
e
lo
ca
l
p
atter
n
s
in
l
o
an
d
ata
a
n
d
R
F
’
s
ca
p
ac
ity
to
h
a
n
d
le
f
ea
tu
r
e
in
ter
ac
tio
n
s
an
d
r
e
d
u
ce
o
v
er
f
itti
n
g
,
m
ak
in
g
it
a
v
iab
le
ap
p
r
o
ac
h
to
im
p
r
o
v
e
o
v
er
all
p
er
f
o
r
m
an
ce
.
Fu
r
th
er
m
o
r
e
,
wh
ile
en
s
em
b
le
m
eth
o
d
s
lik
e
R
F
an
d
XGBo
o
s
t
(
XGB)
d
o
m
i
n
ate
th
is
f
ield
,
t
h
ey
o
f
ten
r
ely
o
n
g
lo
b
al
p
atter
n
s
an
d
m
ay
o
v
er
lo
o
k
lo
ca
l
n
u
an
ce
s
in
lo
an
d
ata.
A
h
y
b
r
id
ap
p
r
o
a
ch
co
m
b
in
in
g
KNN
with
R
F
ca
n
ca
p
tu
r
e
b
o
th
lo
ca
l
p
atter
n
s
(
v
ia
KNN
’
s
in
s
ta
n
ce
-
b
ased
lea
r
n
in
g
)
an
d
g
lo
b
al
in
ter
ac
tio
n
s
(
v
ia
RF
’
s
en
s
em
b
le
ap
p
r
o
ac
h
)
,
p
r
o
v
id
in
g
a
m
o
r
e
co
m
p
r
eh
e
n
s
iv
e
u
n
d
er
s
tan
d
in
g
o
f
l
o
an
d
ef
au
lt
r
is
k
,
lead
i
n
g
to
im
p
r
o
v
e
d
m
o
d
el
’
s
r
o
b
u
s
tn
ess
.
E
x
is
tin
g
s
tu
d
ies
in
clu
d
in
g
th
o
s
e
b
y
[
9
]
,
[
1
0
]
,
o
f
ten
s
u
f
f
e
r
f
r
o
m
m
an
u
al
f
ea
tu
r
e
b
iases
an
d
lack
h
y
p
er
p
ar
am
eter
t
u
n
in
g
.
As
a
r
esu
lt,
th
is
s
tu
d
y
in
tr
o
d
u
ce
s
th
e
co
r
r
elatio
n
-
b
ased
f
ea
t
u
r
e
s
el
ec
tio
n
(
C
B
FS
)
an
d
g
r
id
s
ea
r
ch
cr
o
s
s
-
v
alid
atio
n
(
GSC
V)
to
p
er
f
o
r
m
im
p
o
r
ta
n
t
f
ea
tu
r
e
s
elec
tio
n
an
d
tu
n
e
KNN
an
d
R
F
’
s
h
y
p
er
p
ar
am
et
er
s
,
r
esp
ec
tiv
ely
,
lead
in
g
to
im
p
r
o
v
e
d
m
o
d
el.
T
h
is
s
tu
d
y
ad
v
an
ce
s
th
e
ex
i
s
tin
g
liter
atu
r
e
b
y
h
y
b
r
id
izin
g
KNN
an
d
R
F,
wh
ile
in
teg
r
atin
g
C
B
FS
an
d
GS
C
V
to
b
u
ild
a
lo
an
d
ef
au
lt
p
r
e
d
ictio
n
m
o
d
el
.
T
h
is
ap
p
r
o
ac
h
en
ab
les
m
o
r
e
ac
cu
r
a
te
p
r
ed
ictio
n
s
,
en
h
an
ce
s
o
v
er
a
ll
p
r
ed
ictiv
e
p
er
f
o
r
m
a
n
ce
,
r
ed
u
ce
s
f
in
an
cial
r
is
k
,
an
d
o
f
f
er
s
an
in
n
o
v
ativ
e
f
r
am
ewo
r
k
with
th
e
p
o
ten
tial
to
t
r
an
s
f
o
r
m
lo
a
n
d
ef
au
lt
p
r
ed
ictio
n
an
d
r
is
k
ass
es
s
m
en
t in
th
e
f
in
an
cial
s
ec
to
r
.
Du
e
to
th
e
s
u
r
g
e
o
f
cr
ed
it
r
is
k
,
f
in
a
n
cial
d
is
tr
ess
,
an
d
r
ep
u
tatio
n
al
d
am
ag
e
ca
u
s
ed
b
y
lo
an
d
ef
a
u
lt
[
1
1
]
,
n
u
m
er
o
u
s
s
tu
d
ies ap
p
ly
m
ac
h
in
e
lear
n
in
g
f
o
r
l
o
an
d
ef
au
lt p
r
ed
ictio
n
.
Fo
r
i
n
s
tan
ce
,
R
F o
u
tp
er
f
o
r
m
ed
D
T
at
8
0
%
v
s
.
7
3
%
ac
cu
r
ac
y
[
1
2
]
,
r
ea
ch
ed
9
0
%
ac
cu
r
ac
y
o
n
C
h
in
ese
P2
P
d
ata,
an
d
9
3
%
ac
cu
r
ac
y
,
9
0
%
p
r
ec
is
io
n
,
8
9
% r
ec
all
[
1
3
]
.
R
F a
ls
o
b
ea
t
ar
tific
ial
n
eu
r
al
n
et
wo
r
k
(
ANN)
with
9
5
% a
cc
u
r
a
cy
,
9
0
% r
ec
all,
8
7
%
F1
-
s
co
r
e
[
8
]
.
On
th
e
o
th
er
h
an
d
,
KNN
ac
h
iev
ed
9
8
.
3
0
%
ac
cu
r
ac
y
with
lo
g
is
tic
r
eg
r
ess
io
n
[
1
4
]
an
d
8
8
.
8
9
%
o
u
tp
er
f
o
r
m
in
g
R
F
(
8
4
.
4
4
%)
[
9
]
.
Stack
in
g
m
o
d
els
lik
e
co
n
v
o
l
u
taio
n
l
n
eu
r
al
n
etwo
r
k
-
o
p
tim
ized
(
0
.
9
2
3
8
ac
cu
r
ac
y
,
0
.
9
1
4
7
a
r
ea
u
n
d
er
cu
r
v
e
(
AUC
)
)
in
[
1
5
]
,
li
g
h
t
g
r
ad
ien
t
b
o
o
s
tin
g
m
eth
o
d
(
L
GB
M)
+L
R
(
0
.
7
7
2
AUC)
in
[
1
6
]
,
L
GB
M
(
0
.
7
3
AUC)
[
1
7
]
,
an
d
R
F+GB+
ex
tr
em
e
g
r
a
d
ien
t
b
o
o
s
tin
g
(
XGB)
(
u
p
to
0
.
9
4
4
AUC)
in
[
1
8
]
,
XGB
(
8
1
.
6
7
%
ac
cu
r
a
cy
[
1
9
]
;
9
0
%
ac
cu
r
ac
y
[
2
0
]
)
,
g
r
ad
ien
t
b
o
o
s
tin
g
(
GB
)
(
8
2
.
4
9
%
ac
cu
r
ac
y
[
2
1
]
;
0
.
9
2
4
AUC
[
2
2
]
)
,
L
GB
M+
L
STM
(
8
6
.
8
9
%
ac
cu
r
ac
y
)
[
2
3
]
,
XGB+L
GB
M
(
9
9
.
8
%
a
cc
u
r
ac
y
)
[
2
4
]
,
an
d
GB
+X
G
B
+R
F (
8
6
.
2
3
%
ac
cu
r
ac
y
)
[
2
5
]
s
h
o
wed
g
ain
s
.
C
las
s
ical
an
d
en
s
em
b
le
m
ac
h
i
n
e
lear
n
in
g
alg
o
r
it
h
m
s
l
ik
e
K
NN
an
d
R
F,
r
esp
ec
tiv
ely
d
o
m
in
ate
lo
an
d
ef
au
lt
p
r
ed
ictio
n
b
u
t
s
u
f
f
e
r
f
r
o
m
o
v
er
f
itti
n
g
,
wea
k
lo
ca
l
p
atter
n
ca
p
tu
r
e
,
an
d
i
n
s
u
f
f
icien
t
f
ea
tu
r
e
s
elec
tio
n
/h
y
p
er
p
ar
a
m
eter
tu
n
in
g
.
E
x
is
tin
g
s
tu
d
ies
o
v
er
l
o
o
k
co
m
p
ar
is
o
n
o
f
KNN
an
d
R
F.
T
h
is
s
tu
d
y
f
ills
th
ese
g
ap
s
with
a
co
m
p
ar
ativ
e
s
tu
d
y
o
f
R
F,
KN
N,
an
d
h
y
b
r
i
d
m
eth
o
d
s
,
co
m
b
in
i
n
g
p
ar
am
e
ter
o
p
tim
izatio
n
to
en
h
an
ce
p
er
f
o
r
m
an
ce
a
n
d
en
s
u
r
e
r
o
b
u
s
t p
r
ed
ictio
n
r
esu
lt.
2.
M
E
T
H
O
D
T
h
i
s
s
ec
t
io
n
o
u
tl
in
e
s
th
e
s
t
ag
es
in
d
ev
el
o
p
in
g
th
e
p
r
o
p
o
s
e
d
co
m
p
ar
at
iv
e
s
tu
d
y
o
f
K
NN
,
R
F
an
d
h
y
b
r
id
m
e
th
o
d
s
f
o
r
l
o
an
d
ef
a
u
lt
p
r
ed
i
ct
io
n
,
i
n
t
eg
r
a
tin
g
C
B
F
S
f
o
r
f
ea
tu
r
e
s
e
le
ct
io
n
an
d
GS
C
V
an
d
r
a
n
d
o
m
s
e
ar
ch
cr
o
s
s
-
v
al
id
a
t
io
n
(
R
S
C
V
)
f
o
r
p
ar
am
et
er
o
p
tim
izatio
n
to
b
o
o
s
t
ac
cu
r
a
cy
a
n
d
r
o
b
u
s
tn
e
s
s
in
f
in
a
n
c
ial
r
i
s
k
a
s
s
e
s
s
m
en
t
,
a
s
i
llu
s
tr
a
ted
in
Fig
u
r
e
1
.
2
.
1
.
Da
t
a
a
cquis
it
io
n a
nd
pr
epro
ce
s
s
ing
I
n
th
is
p
ap
er
,
a
lo
an
d
ef
au
lt
d
ataset
is
ac
q
u
ir
ed
f
r
o
m
th
e
Ka
g
g
le
r
ep
o
s
ito
r
y
(
h
ttp
s
://www.
k
ag
g
le.
co
m
/d
ata
s
ets/
y
as
s
er
h
/lo
an
-
d
ef
au
lt
-
d
ataset
),
co
n
tain
in
g
1
48
,
6
7
0
i
n
s
tan
ce
s
an
d
3
4
f
ea
tu
r
es,
in
clu
d
in
g
th
e
ta
r
g
et
f
ea
tu
r
e.
T
h
e
d
ataset
u
n
d
er
wen
t
v
ar
io
u
s
p
r
e
p
r
o
ce
s
s
in
g
s
tag
es
in
clu
d
in
g
m
is
s
in
g
v
alu
es
tr
ea
tm
en
t
u
s
in
g
s
im
p
l
e
im
p
u
ter
,
ca
teg
o
r
ical
v
ar
ia
b
l
e
en
co
d
in
g
u
s
in
g
la
b
el
en
co
d
er
,
tr
an
s
f
o
r
m
atio
n
u
s
in
g
s
tan
d
ar
d
s
ca
ler
,
a
n
d
cla
s
s
b
alan
cin
g
u
s
in
g
s
y
n
th
etic
m
in
o
r
ity
o
v
e
r
s
am
p
lin
g
tech
n
i
q
u
e
(
SMOT
E
)
,
as
a
r
esu
lt
o
f
a
n
im
b
alan
ce
d
clas
s
d
is
tr
ib
u
tio
n
with
7
5
.
3
6
%
o
f
in
s
tan
ce
s
b
elo
n
g
in
g
to
class
0
an
d
2
4
.
6
4
%
b
elo
n
g
in
g
to
class
1
as sh
o
wn
in
Fig
u
r
e
2
.
Evaluation Warning : The document was created with Spire.PDF for Python.
C
o
m
p
u
t Sci
I
n
f
T
ec
h
n
o
l
I
SS
N:
2722
-
3
2
2
1
A
co
mp
a
r
a
tive
s
tu
d
y
o
f c
l
a
s
s
ica
l,
b
a
g
g
in
g
,
a
n
d
h
yb
r
id
meth
o
d
s
fo
r
o
p
timiz
in
g
lo
a
n
…
(
I
s
ma
il I
d
o
w
u
A
ku
ji
)
181
Fig
u
r
e
1
.
Mo
d
el
d
ev
elo
p
m
en
t
p
r
o
ce
s
s
Af
ter
class
b
alan
cin
g
,
th
e
class
d
is
tr
ib
u
tio
n
b
ec
o
m
es
5
0
%
o
f
in
s
tan
ce
s
in
ea
ch
class
as
s
h
o
wn
in
Fig
u
r
e
3
.
T
h
is
class
b
alan
cin
g
o
f
f
e
r
s
r
ed
u
ce
d
b
ias
to
war
d
s
n
o
n
-
d
e
f
au
lt
ca
s
es
b
ec
au
s
e
t
h
e
m
o
d
el
u
s
in
g
t
h
e
o
r
ig
in
al
d
ataset
is
less
lik
ely
to
b
e
b
iased
to
war
d
s
p
r
e
d
ictin
g
n
o
n
-
d
e
f
au
lt
ca
s
es
(
cla
s
s
0
)
,
wh
ich
wer
e
p
r
ev
io
u
s
ly
t
h
e
m
ajo
r
ity
class
.
T
h
e
b
alan
ce
d
d
ataset
en
ab
les
a
m
o
r
e
ac
c
u
r
ate
ass
ess
m
en
t
o
f
th
e
r
is
k
ass
o
ciate
d
with
lo
an
d
ef
a
u
lt,
en
a
b
lin
g
le
n
d
er
s
to
m
ak
e
m
o
r
e
i
n
f
o
r
m
e
d
d
ec
is
io
n
s
.
I
n
ad
d
itio
n
,
s
ev
e
n
f
ea
tu
r
es
with
th
e
h
ig
h
est
im
p
o
r
tan
t
s
co
r
es
w
er
e
s
elec
ted
u
s
in
g
C
B
FS
t
ec
h
n
iq
u
e
,
s
p
ec
if
ically
u
s
in
g
SelectKBe
s
t
with
m
u
tu
al_
in
f
o
_
class
if
as th
e
s
co
r
in
g
f
u
n
ctio
n
.
Fig
u
r
e
2
.
Dis
tr
ib
u
tio
n
o
f
tar
g
e
t f
ea
tu
r
e
b
ef
o
r
e
class
b
alan
cin
g
Fig
u
r
e
3
.
Dis
tr
ib
u
tio
n
o
f
tar
g
e
t f
ea
tu
r
e
af
ter
class
b
alan
cin
g
2
.
2
.
M
o
del
dev
elo
pm
ent
us
ing
K
NN
KN
N
i
s
a
n
o
n
-
p
ar
am
etr
ic
,
s
u
p
er
v
i
s
ed
m
ac
h
in
e
le
ar
n
in
g
al
g
o
r
it
h
m
th
at
c
la
s
s
i
f
i
es
a
n
ew
d
a
ta
p
o
in
t
b
y
id
e
n
t
if
y
i
n
g
i
t
i
s
k
clo
s
e
s
t
n
eig
h
b
o
r
s
in
th
e
tr
ai
n
in
g
d
a
t
a
b
a
s
ed
o
n
a
d
i
s
tan
ce
m
et
r
i
c
an
d
a
s
s
ig
n
in
g
th
e
m
ajo
r
i
ty
c
la
s
s
lab
el
am
o
n
g
th
em
.
T
h
e
p
s
eu
d
o
co
d
e
o
f
KNN
is
s
h
o
wn
in
Alg
o
r
ith
m
1
.
Alg
o
rit
hm
1
.
P
s
eudo
co
de
f
o
r
im
plem
ent
ing
k
-
nea
re
s
t
nei
g
hb
o
r
Giv
en
th
e
tr
ain
in
g
d
ataset: {
(
x
1
, y
1
)
,
(
x
2
, y
2
)
,
……
,
(
x
m
,
y
m
)}
1
.
Sto
r
e
th
e
tr
ai
n
in
g
s
et
2
.
Fo
r
ea
ch
n
ew
u
n
lab
eled
d
at
a,
a
.
C
alcu
late
E
u
clid
ea
n
d
is
tan
c
e
with
all
tr
ain
in
g
d
ata
p
o
in
ts
u
s
in
g
th
e
f
o
r
m
u
la:
√
∑
(
=
1
−
)
2
b
.
Fin
d
th
e
k
-
n
ea
r
est n
eig
h
b
o
u
r
s
c
.
Ass
ig
n
class
co
n
tain
in
g
th
e
m
ax
im
u
m
n
u
m
b
e
r
o
f
n
ea
r
est n
eig
h
b
o
u
r
s
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
7
2
2
-
3
2
2
1
C
o
m
p
u
t Sci
I
n
f
T
ec
h
n
o
l
,
Vo
l.
7
,
No
.
2
,
J
u
ly
20
26
:
179
-
1
95
182
R
F
is
an
en
s
em
b
le
lear
n
in
g
alg
o
r
ith
m
th
at
co
m
b
in
es
m
u
lti
p
le
d
ec
is
io
n
tr
ee
s
to
im
p
r
o
v
e
p
r
ed
ictio
n
ac
cu
r
ac
y
an
d
r
ed
u
ce
o
v
er
f
itti
n
g
.
I
n
th
is
ar
ticle,
n
_
esti
m
ato
r
s
(
n
u
m
b
er
o
f
tr
ee
s
)
ar
e
s
et
to
1
0
0
with
r
an
d
o
m
_
s
tate
s
et
to
4
2
to
e
n
s
u
r
e
r
ep
r
o
d
u
cib
ilit
y
.
T
h
e
p
s
eu
d
o
co
d
e
o
f
R
F is
s
h
o
wn
in
Alg
o
r
i
th
m
2
.
Alg
o
rit
hm
2
.
P
s
eudo
co
de
f
o
r
im
plem
ent
ing
ra
nd
o
m
f
o
re
s
t
1
.
I
n
p
u
t:
-
T
r
ai
n
in
g
d
ataset
D=
{(
x
1
,
y
1
)
,
(
x
2
,
y
2
)
,
.
.
.
,
(
x
n
,
y
n
)
}
-
Nu
m
b
er
o
f
tr
ee
s
T
-
Nu
m
b
er
o
f
f
ea
tu
r
es m
2
.
Fo
r
t=1
t
o
T
:
-
B
o
o
ts
tr
ap
s
am
p
lin
g
: Cre
ate
a
b
o
o
ts
tr
ap
s
am
p
le
Dt
f
r
o
m
D
-
Gr
o
w
a
tr
ee
:
-
Fo
r
ea
ch
n
o
d
e:
-
R
an
d
o
m
ly
s
elec
t m
f
ea
tu
r
es
-
Fin
d
th
e
b
est s
p
lit f
ea
tu
r
e
an
d
v
alu
e
-
Sp
lit th
e
n
o
d
e
in
to
tw
o
ch
ild
n
o
d
es
-
S
to
r
e
th
e
tr
ee
3
.
Ou
tp
u
t: E
n
s
em
b
le
o
f
T
tr
ee
s
.
T
h
e
m
o
d
els
wer
e
f
in
etu
n
ed
u
s
in
g
GSC
V
an
d
R
S
C
V
tec
h
n
iq
u
es.
T
h
e
h
y
p
e
r
p
ar
am
eter
g
r
id
s
f
o
r
tu
n
in
g
wer
e
d
e
f
in
ed
as
f
o
llo
w
s
:
f
o
r
KNN,
th
e
n
u
m
b
er
o
f
n
e
ig
h
b
o
r
s
(
n
_
n
eig
h
b
o
r
s
)
was
tu
n
ed
o
v
e
r
[
3
,
5
,
7
,
9
,
1
1
]
,
cv
=5
;
f
o
r
R
F,
th
e
n
u
m
b
er
o
f
esti
m
ato
r
s
(
n
_
esti
m
ato
r
s
)
was
tu
n
ed
o
v
er
[
5
0
,
1
0
0
,
2
0
0
]
,
cv
=
5
an
d
th
e
m
ax
im
u
m
d
e
p
th
(
m
a
x
_
d
e
p
th
)
was
tu
n
ed
o
v
er
[
No
n
e,
5
,
1
0
]
;
f
o
r
th
e
Hy
b
r
id
m
o
d
el,
b
o
t
h
n
_
n
ei
g
h
b
o
r
s
an
d
n
_
esti
m
ato
r
s
wer
e
tu
n
ed
o
v
er
th
e
s
am
e
r
an
g
es a
s
KNN
an
d
R
F,
r
esp
ec
tiv
ely
.
2
.
3
.
P
er
f
o
r
m
a
nce
ev
a
lua
t
io
n
o
f
t
he
dev
elo
ped mo
dels
T
h
e
ev
alu
atio
n
m
etr
ics
u
s
ed
in
clu
d
e
ac
cu
r
ac
y
,
p
r
ec
is
io
n
,
r
ec
all,
f
1
-
s
co
r
e,
a
n
d
r
ec
ei
v
er
o
p
er
atin
g
ch
ar
ac
ter
is
tics
(
R
OC
)
cu
r
v
e
m
etr
ics.
Acc
u
r
ac
y
ca
lcu
lates
th
e
p
r
o
p
o
r
tio
n
o
f
co
r
r
ec
tly
p
r
ed
icted
o
u
tco
m
es,
ex
p
r
ess
ed
as
a
p
er
ce
n
tag
e,
u
s
in
g
a
s
p
ec
if
ic
f
o
r
m
u
la
to
q
u
an
tify
its
v
alu
e.
Acc
u
r
ac
y
is
m
ath
em
atica
lly
r
ep
r
esen
ted
as f
o
llo
ws:
=
+
+
+
+
(
1
)
Mo
r
eo
v
er
,
p
r
ec
is
io
n
ca
lcu
late
s
th
e
p
r
o
p
o
r
tio
n
o
f
c
o
r
r
ec
tly
p
r
ed
icted
p
o
s
itiv
e
in
s
tan
ce
s
o
u
t
o
f
all
p
r
ed
icted
p
o
s
itiv
e
in
s
tan
ce
s
,
p
r
o
v
id
in
g
a
m
ea
s
u
r
e
o
f
e
x
ac
tn
e
s
s
.
I
t c
an
b
e
d
eter
m
i
n
ed
v
ia:
=
+
(
2
)
R
ec
all
m
ea
s
u
r
es th
e
m
o
d
el
’
s
ab
ilit
y
to
d
etec
t
ac
tu
al
lo
an
d
ef
au
lts
.
Ma
th
em
atica
lly
,
R
e
c
a
l
l
=
TP
TP
+
FN
(
3
)
F1
-
Sco
r
e
b
alan
ce
s
p
r
ec
is
io
n
a
n
d
r
ec
all.
Ma
th
em
atica
lly
,
F1
−
s
c
or
e
=
2
×
×
+
(
4
)
W
h
er
e,
T
P,
T
N,
FN
,
an
d
FP
d
en
o
te
tr
u
e
p
o
s
itiv
e,
tr
u
e
n
e
g
ati
v
e,
f
alse n
eg
ativ
e
a
n
d
f
alse p
o
s
itiv
e,
r
esp
ec
tiv
ely
.
L
astl
y
,
th
e
R
O
C
cu
r
v
e
will
b
e
p
lo
tted
u
s
in
g
th
e
tr
u
e
p
o
s
itiv
e
r
ate
(
T
PR
)
ag
ain
s
t
th
e
f
alse
p
o
s
itiv
e
r
ate
(
FP
R
)
at
d
if
f
er
en
t th
r
esh
o
ld
s
.
3.
RE
SU
L
T
S AN
D
D
I
SCU
SS
I
O
N
3
.
1
.
Resul
t
s
o
f
da
t
a
prepro
ce
s
s
ing
T
h
e
lo
an
d
ataset
co
n
tain
s
n
o
n
-
ca
teg
o
r
ical
f
ea
tu
r
es,
m
ak
in
g
it
im
p
e
r
ativ
e
to
p
r
e
p
r
o
ce
s
s
it
to
b
e
s
u
itab
le
f
o
r
m
o
d
el
d
e
v
elo
p
m
e
n
t.
Hen
ce
,
th
is
s
tu
d
y
em
p
l
o
y
e
d
L
ab
el
E
n
co
d
er
to
tr
a
n
s
f
o
r
m
th
e
d
ataset
i
n
to
a
s
u
itab
le
f
o
r
m
at.
T
h
e
en
c
o
d
e
d
f
ea
tu
r
es,
alo
n
g
with
t
h
eir
co
r
r
esp
o
n
d
in
g
p
r
o
ce
s
s
v
alu
es
,
ar
e
v
is
u
alize
d
in
Fig
u
r
e
4
,
s
h
o
wca
s
in
g
th
e
tr
an
s
f
o
r
m
atio
n
o
u
tco
m
e.
Evaluation Warning : The document was created with Spire.PDF for Python.
C
o
m
p
u
t Sci
I
n
f
T
ec
h
n
o
l
I
SS
N:
2722
-
3
2
2
1
A
co
mp
a
r
a
tive
s
tu
d
y
o
f c
l
a
s
s
ica
l,
b
a
g
g
in
g
,
a
n
d
h
yb
r
id
meth
o
d
s
fo
r
o
p
timiz
in
g
lo
a
n
…
(
I
s
ma
il I
d
o
w
u
A
ku
ji
)
183
Fig
u
r
e
4
.
L
a
b
el
en
co
d
in
g
o
f
th
e
ca
teg
o
r
ical
v
ar
ia
b
les an
d
th
e
ir
co
r
r
esp
o
n
d
in
g
v
alu
es
Han
d
lin
g
m
is
s
in
g
v
alu
es
is
a
c
r
u
cial
asp
ec
t
o
f
d
ata
p
r
ep
r
o
ce
s
s
in
g
b
ec
au
s
e
it
ca
n
s
ig
n
if
ican
t
ly
im
p
ac
t
th
e
p
er
f
o
r
m
an
ce
a
n
d
r
eliab
ilit
y
o
f
m
ac
h
in
e
lear
n
i
n
g
m
o
d
e
ls
,
p
o
ten
tially
lead
in
g
t
o
b
iased
r
esu
lts
,
r
ed
u
ce
d
m
o
d
el
p
er
f
o
r
m
a
n
ce
,
an
d
lo
s
s
o
f
v
alu
a
b
le
in
f
o
r
m
a
tio
n
if
n
o
t
p
r
o
p
er
l
y
ad
d
r
ess
ed
.
T
h
e
s
im
p
l
e
im
p
u
ter
tech
n
iq
u
e
is
u
s
ed
in
th
is
s
tu
d
y
to
h
a
n
d
le
m
is
s
in
g
v
alu
e
an
d
t
h
e
p
r
o
ce
s
s
an
d
r
esu
lt a
r
e
s
h
o
wn
in
T
a
b
le
1
.
Ad
d
itio
n
ally
,
as
p
ar
t
o
f
th
e
d
ata
p
r
ep
r
o
ce
s
s
in
g
is
th
e
s
elec
t
io
n
o
f
im
p
o
r
ta
n
t
f
ea
tu
r
es.
I
n
th
is
s
tu
d
y
,
s
ev
en
f
ea
tu
r
es
m
et
th
e
cr
iter
ia
s
et
(
th
r
esh
o
ld
>=
0
.
2
5
)
f
o
r
th
e
s
elec
tio
n
o
f
im
p
o
r
tan
t
f
e
atu
r
e
s
u
s
in
g
C
B
FS
.
T
h
e
s
elec
ted
f
ea
tu
r
es,
s
h
o
wn
in
Fig
u
r
e
5
ar
e
f
o
u
n
d
t
o
h
a
v
e
th
e
h
ig
h
est
c
o
r
r
elatio
n
co
ef
f
i
cien
t
to
th
e
tar
g
et
v
ar
iab
le.
T
h
e
f
ea
tu
r
es
ar
e:
‘
Up
f
r
o
n
t
_
ch
ar
g
es
’
,
‘
I
n
ter
est_
r
ate_
s
p
r
ea
d
’
,
‘
p
r
o
p
er
ty
_
v
alu
e
’
,
‘
co
-
ap
p
lican
t_
cr
ed
it
_
ty
p
e
’
,
‘
lu
m
p
_
s
u
m
_
p
ay
m
en
t
’
,
‘
Neg
_
am
m
o
r
tizatio
n
’
,
an
d
‘
s
u
b
m
is
s
io
n
_
o
f
_
ap
p
licatio
n
’
.
T
ab
le
1
.
R
esu
lts
o
f
s
im
p
le
im
p
u
ter
f
o
r
m
is
s
in
g
v
alu
es tr
ea
tm
en
t
F
e
a
t
u
r
e
s
B
e
f
o
r
e
i
mp
u
t
a
t
i
o
n
A
f
t
e
r
i
m
p
u
t
a
t
i
o
n
S
u
m
o
f
mi
ssi
n
g
v
a
l
u
e
s
N
u
mb
e
r
o
f
mi
ss
i
n
g
v
a
l
u
e
s
r
a
t
e
_
o
f
_
i
n
t
e
r
e
s
t
3
6
4
3
9
1
1
2
2
3
1
36439
112231
0
I
n
t
e
r
e
st
_
r
a
t
e
_
sp
r
e
a
d
3
6
6
3
9
1
1
2
0
3
1
36639
112031
0
U
p
f
r
o
n
t
_
c
h
a
r
g
e
s
3
9
6
4
2
1
0
9
0
2
8
39642
109028
0
t
e
r
m
41
1
4
8
6
2
9
41
148629
0
p
r
o
p
e
r
t
y
_
v
a
l
u
e
1
5
0
9
8
1
3
3
5
7
2
15098
133572
0
i
n
c
o
m
e
9
1
5
0
1
3
9
5
2
0
9150
139520
0
LTV
1
5
0
9
8
1
3
3
5
7
2
15098
133572
0
d
t
i
r
1
2
4
1
2
1
1
2
4
5
4
9
24121
124549
0
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
7
2
2
-
3
2
2
1
C
o
m
p
u
t Sci
I
n
f
T
ec
h
n
o
l
,
Vo
l.
7
,
No
.
2
,
J
u
ly
20
26
:
179
-
1
95
184
Fig
u
r
e
5
.
Featu
r
es
im
p
o
r
tan
ce
s
co
r
es o
f
th
e
s
elec
ted
f
ea
tu
r
es
3
.
2
.
Resul
t
s
o
f
m
o
dels
’
perf
o
rm
a
nce
ev
a
lua
t
io
n wit
ho
ut
hy
perpa
ra
m
et
er
t
un
ing
3
.
2
.
1
.
Resul
t
s
o
f
k
-
nea
re
s
t
ne
ig
hb
o
r
m
o
del
T
h
e
co
n
f
u
s
io
n
m
at
r
ix
s
h
o
wn
in
Fig
u
r
e
6
s
h
o
ws
th
at
th
e
KNN
ac
h
iev
ed
:
T
P=7
3
1
5
,
T
N=
2
2
1
1
8
,
FP
=2
8
8
,
FN=1
3
,
r
e
p
r
esen
tin
g
2
9
,
4
3
3
c
o
r
r
ec
t
class
if
icatio
n
s
an
d
3
0
1
m
is
class
if
icatio
n
s
.
B
ased
o
n
th
e
co
n
f
u
s
io
n
m
atr
ix
in
Fig
u
r
e
6
,
t
h
e
f
o
llo
win
g
m
etr
ics ca
lcu
latio
n
s
ca
n
b
e
d
er
iv
e
d
:
:
+
+
+
+
=
7315
+
2
94
3
3
7315
+
2
94
3
3
+
288
+
13
≈
0
.
992
:
+
=
7315
7315
+
288
≈
0
.
962
:
+
=
7315
7315
+
13
≈
0
.
998
1
−
:
2
×
×
+
=
2
×
0
.
96
2
×
0
.
9
98
0
.
96
2
+
0
.
998
≈
0
.
980
Fig
u
r
e
6
.
C
o
n
f
u
s
io
n
m
atr
i
x
o
f
th
e
KNN
m
o
d
el
r
esu
lt
Evaluation Warning : The document was created with Spire.PDF for Python.
C
o
m
p
u
t Sci
I
n
f
T
ec
h
n
o
l
I
SS
N:
2722
-
3
2
2
1
A
co
mp
a
r
a
tive
s
tu
d
y
o
f c
l
a
s
s
ica
l,
b
a
g
g
in
g
,
a
n
d
h
yb
r
id
meth
o
d
s
fo
r
o
p
timiz
in
g
lo
a
n
…
(
I
s
ma
il I
d
o
w
u
A
ku
ji
)
185
3
.
2
.
2
.
Resul
t
s
o
f
ra
nd
o
m
f
o
r
est
m
o
del
T
h
e
co
n
f
u
s
io
n
m
atr
ix
s
h
o
wn
i
n
Fig
u
r
e
7
s
h
o
ws
th
at
th
e
R
F
ac
h
iev
ed
:
T
P=7
3
2
6
,
T
N=
2
2
3
9
4
,
FP
=1
2
,
FN=2
,
r
ep
r
esen
tin
g
2
9
,
7
2
0
c
o
r
r
ec
t
class
if
icatio
n
s
an
d
1
4
m
i
s
class
if
icatio
n
s
.
B
ased
o
n
th
e
co
n
f
u
s
io
n
m
atr
ix
in
Fig
u
r
e
7
,
th
e
f
o
llo
win
g
m
etr
ic
s
ca
lcu
latio
n
s
ca
n
b
e
d
er
i
v
ed
:
:
+
+
+
+
=
7326
+
2
2
3
94
7326
+
2
2
3
94
+
12
+
2
≈
1
.
000
:
+
=
7326
7326
+
12
≈
0
.
998
:
:
+
=
7326
7326
+
2
≈
1
.
000
1
−
:
2
×
×
+
=
2
×
0
.
998
×
1
.
000
0
.
998
+
1
.
000
≈
0
.
999
Fig
u
r
e
7
.
C
o
n
f
u
s
io
n
m
atr
i
x
o
f
th
e
R
F
m
o
d
el
r
esu
lt
3
.
2
.
3
.
Resul
t
s
o
f
t
he
hy
brid
K
NN
-
RF
m
o
del
T
h
e
co
n
f
u
s
io
n
m
atr
i
x
s
h
o
wn
in
Fig
u
r
e
8
s
h
o
ws
th
at
th
e
h
y
b
r
id
KNN+
R
F
ac
h
iev
ed
:
T
P=7
3
2
4
,
T
N=
2
2
3
4
2
,
FP
=6
4
,
FN=4
,
r
ep
r
esen
tin
g
2
9
,
6
6
6
co
r
r
ec
t c
lass
i
f
icatio
n
s
an
d
6
8
m
is
class
if
icat
io
n
s
.
Fig
u
r
e
8
.
C
o
n
f
u
s
io
n
m
atr
i
x
o
f
th
e
h
y
b
r
id
KNN+
R
F
m
o
d
el
r
e
s
u
lt
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
7
2
2
-
3
2
2
1
C
o
m
p
u
t Sci
I
n
f
T
ec
h
n
o
l
,
Vo
l.
7
,
No
.
2
,
J
u
ly
20
26
:
179
-
1
95
186
B
ased
o
n
th
e
co
n
f
u
s
io
n
m
atr
i
x
in
Fig
u
r
e
8
,
th
e
f
o
llo
win
g
m
et
r
ics ca
lcu
latio
n
s
ca
n
b
e
d
e
r
iv
e
d
:
:
+
+
+
+
=
7324
+
2
2
3
4
2
7324
+
2
2
3
4
2
+
64
+
4
≈
0
.
998
:
+
=
7324
7324
+
64
≈
0
.
991
:
+
=
7324
7324
+
4
≈
0
.
999
1
−
:
2
×
×
+
=
2
×
0
.
991
×
0
.
999
0
.
991
+
0
.
999
≈
0
.
995
Fu
r
th
er
m
o
r
e
,
th
e
AUC
-
R
O
C
s
co
r
e
o
f
th
e
KNN
r
esu
lted
in
0
.
9
9
7
,
w
h
ile
R
F
an
d
h
y
b
r
i
d
KNN+
R
F
r
esu
lted
in
1
.
0
0
0
,
in
d
icatin
g
s
tr
o
n
g
m
o
d
el
p
er
f
o
r
m
a
n
ce
.
T
h
e
R
OC
cu
r
v
e
is
r
ep
r
esen
ted
in
Fig
u
r
e
9
.
T
a
b
le
2
co
m
p
ar
es
t
h
e
p
er
f
o
r
m
an
ce
o
f
th
e
th
r
ee
m
ac
h
in
e
lear
n
in
g
m
o
d
els
-
KNN,
R
F,
an
d
Hy
b
r
id
KNN+
R
F
-
ac
r
o
s
s
v
ar
io
u
s
m
etr
ics.
T
h
e
R
F
m
o
d
el
o
u
tp
er
f
o
r
m
s
th
e
o
th
er
s
wit
h
1
0
0
%
ac
cu
r
ac
y
,
p
r
ec
is
io
n
,
r
ec
all,
an
d
F1
-
s
co
r
e,
an
d
th
e
h
ig
h
est
R
OC
-
AU
C
s
c
o
r
e.
Ho
wev
er
,
it
i
s
th
e
KNN
m
o
d
el
th
at
i
s
th
e
s
lo
west
to
tr
ain
(
5
1
.
5
4
s
ec
o
n
d
s
)
,
wh
ile
th
e
R
F
m
o
d
el
is
th
e
f
as
t
est
(
0
.
9
4
s
ec
o
n
d
s
)
.
T
h
e
h
y
b
r
id
m
o
d
el
b
alan
ce
s
p
er
f
o
r
m
an
ce
an
d
s
p
ee
d
b
u
t
d
o
es
n
o
t
s
u
r
p
ass
th
e
R
F
m
o
d
el
’
s
a
cc
u
r
ac
y
.
Ov
e
r
all,
th
e
R
F
m
o
d
el
is
th
e
b
est
ch
o
ice
f
o
r
th
is
task
d
u
e
t
o
it
i
s
h
ig
h
p
er
f
o
r
m
an
ce
an
d
f
ast
tr
ain
in
g
tim
e.
T
h
e
h
y
b
r
id
m
o
d
el
co
u
ld
b
e
an
alter
n
ati
v
e,
o
f
f
er
in
g
a
b
alan
ce
b
etwe
e
n
KNN
an
d
R
F st
r
en
g
th
s
.
Fig
u
r
e
9
.
R
OC
cu
r
v
e
o
f
th
e
K
NN,
R
F,
an
d
h
y
b
r
i
d
m
o
d
el
T
ab
le
2
.
C
o
m
p
a
r
is
o
n
o
f
th
e
K
NN,
R
F a
n
d
h
y
b
r
id
KNN+
R
F
m
o
d
els
M
e
t
r
i
c
K
N
N
RF
H
y
b
r
i
d
K
N
N
+
R
F
A
c
c
u
r
a
c
y
9
9
.
2
%
1
0
0
%
9
9
.
8
%
P
r
e
c
i
s
i
o
n
9
6
.
2
%
9
9
.
8
%
9
9
.
1
%
R
e
c
a
l
l
9
9
.
8
%
1
0
0
%
9
9
.
9
%
F1
-
S
c
o
r
e
9
8
.
0
%
9
9
.
9
%
9
9
.
5
%
R
O
C
-
AUC
0
.
9
9
7
1
.
0
0
0
1
.
0
0
0
Tr
a
i
n
i
n
g
T
i
me
5
1
.
5
4
2
7
6
6
0
.
9
4
1
2
1
8
4
0
.
6
1
2
2
4
1
M
e
m
o
r
y
(
M
b
)
4
3
5
.
7
3
8
2
8
1
4
3
5
.
7
6
1
7
1
9
4
3
5
.
7
7
3
4
3
8
3
.
3
.
Resul
t
s
o
f
m
o
dels
’
perf
o
rm
a
nce
ev
a
lua
t
io
n wit
h t
un
ing
(
G
SCV
a
nd
RSCV)
3
.
3
.
1
.
Resul
t
s
o
f
t
un
ed
m
o
dels
wit
h G
SCV
T
h
e
co
n
f
u
s
io
n
m
atr
ix
s
h
o
w
n
in
Fig
u
r
e
1
0
s
h
o
ws
th
at
th
e
KNN+
GS
C
V
ac
h
iev
ed
:
T
P=7
3
1
7
,
T
N=
2
2
1
7
5
,
FP
=1
1
,
FN=2
3
1
.
Fu
r
th
er
m
o
r
e,
Fig
u
r
e
1
1
r
ev
ea
ls
th
at
R
F+G
SC
V
ac
h
iev
e
d
T
P
(
7
3
2
8
)
,
T
N
(
2
2
3
9
4
)
,
FP
(
1
2
)
,
FN
(
0
)
an
d
in
Fig
u
r
e
1
2
,
h
y
b
r
id
+G
SC
V
ac
co
u
n
ted
f
o
r
T
P
(
7
3
2
3
)
,
T
N
(
2
2
2
8
8
)
,
FP
(
1
1
8
)
,
FN (
5
)
.
Evaluation Warning : The document was created with Spire.PDF for Python.
C
o
m
p
u
t Sci
I
n
f
T
ec
h
n
o
l
I
SS
N:
2722
-
3
2
2
1
A
co
mp
a
r
a
tive
s
tu
d
y
o
f c
l
a
s
s
ica
l,
b
a
g
g
in
g
,
a
n
d
h
yb
r
id
meth
o
d
s
fo
r
o
p
timiz
in
g
lo
a
n
…
(
I
s
ma
il I
d
o
w
u
A
ku
ji
)
187
Fig
u
r
e
1
0
.
C
o
n
f
u
s
io
n
m
atr
ix
o
f
th
e
KNN_
GSC
V
m
o
d
el
r
esu
lt
Fig
u
r
e
1
1
.
C
o
n
f
u
s
io
n
m
atr
ix
o
f
th
e
R
F_
GSC
V
m
o
d
el
r
esu
lt
Fig
u
r
e
1
2
.
C
o
n
f
u
s
io
n
m
atr
ix
o
f
th
e
h
y
b
r
id
_
GSC
V
m
o
d
el
r
esu
lt
T
h
e
m
o
d
els
’
r
esu
lts
with
GSC
V
f
r
o
m
co
n
f
u
s
io
n
m
atr
ices
Fig
u
r
e
1
0
-
1
2
a
r
e
p
r
esen
ted
in
T
ab
le
3
,
with
R
F_
GS
C
V
b
ein
g
th
e
b
est m
o
d
el,
f
o
llo
wed
b
y
Hy
b
r
id
_
GSC
V.
T
ab
le
3
.
C
o
m
p
a
r
is
o
n
o
f
th
e
K
NN,
R
F a
n
d
h
y
b
r
id
with
GSC
V
M
e
t
r
i
c
K
N
N
_
G
S
C
V
TP (
7
3
1
7
)
,
TN
(
2
2
1
7
5
)
,
F
P
(
1
1
)
,
F
N
(
2
3
1
)
R
F
_
G
S
C
V
TP (
7
3
2
8
)
,
TN
(
2
2
3
9
4
)
,
F
P
(
1
2
)
,
F
N
(
0
)
H
y
b
r
i
d
_
G
S
C
V
TP (
7
3
2
3
)
,
TN
(
2
2
2
8
8
)
,
F
P
(
1
1
8
)
,
F
N
(
5
)
A
c
c
u
r
a
c
y
:
+
+
+
+
9
9
.
2
%
1
0
0
%
9
9
.
6
%
P
r
e
c
i
s
i
o
n
:
+
9
9
.
8
%
9
9
.
8
%
9
8
.
4
%
R
e
c
a
l
l
:
+
9
6
.
9
%
1
0
0
%
9
9
.
9
%
F1
-
S
c
o
r
e
:
9
8
.
4
%
9
9
.
9
%
9
9
.
2
%
R
O
C
-
A
U
C
:
×
×
+
0
.
9
9
7
1
.
0
0
0
1
.
0
0
0
Ti
me
2
2
.
1
0
4
9
0
.
5
9
5
4
2
2
.
4
5
1
0
3
.
3
.
2
.
Resul
t
s
o
f
t
un
ed
m
o
dels
wit
h RSCV
Fig
u
r
e
s
1
3
-
1
5
r
e
v
ea
ls
th
e
co
n
f
u
s
io
n
m
atr
ices
o
f
th
e
m
o
d
els
af
ter
tu
n
in
g
with
R
SC
V.
T
h
e
r
esu
lts
ar
e
as
f
o
llo
w:
KNN_
R
SC
V:
T
P
(
7
3
1
7
)
,
T
N
(
2
2
1
7
5
)
,
FP
(
1
1
)
,
FN
(
2
3
1
)
;
R
F_
R
SC
V:
T
P
(
7
3
2
8
)
,
T
N
(
2
2
3
9
4
)
,
FP
(
1
2
)
,
FN (
0
)
; H
y
b
r
id
_
R
SC
V:
T
P (
7
3
2
3
)
,
T
N
(
2
2
2
8
8
)
,
FP
(
1
1
8
)
,
FN (
5
)
.
T
h
e
m
o
d
els
’
r
esu
l
ts
with
RS
C
V
f
r
o
m
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
7
2
2
-
3
2
2
1
C
o
m
p
u
t Sci
I
n
f
T
ec
h
n
o
l
,
Vo
l.
7
,
No
.
2
,
J
u
ly
20
26
:
179
-
1
95
188
co
n
f
u
s
io
n
m
atr
ices
in
Fig
u
r
e
s
1
3
-
1
5
ar
e
p
r
esen
ted
i
n
T
ab
le
4
,
with
R
F_
R
S
C
V
b
ein
g
th
e
b
e
s
t
m
o
d
el,
f
o
llo
we
d
b
y
Hy
b
r
id
_
R
SC
V.
Fig
u
r
e
1
3
.
C
o
n
f
u
s
io
n
m
atr
ix
o
f
th
e
KNN_
R
SC
V
m
o
d
el
r
esu
lt
Fig
u
r
e
1
4
.
C
o
n
f
u
s
io
n
m
atr
ix
o
f
th
e
R
F_
R
SC
V
m
o
d
el
r
esu
lt
Fig
u
r
e
1
5
.
C
o
n
f
u
s
io
n
m
atr
ix
o
f
th
e
h
y
b
r
id
_
R
SC
V
m
o
d
el
r
esu
lt
T
ab
le
4
.
C
o
m
p
a
r
is
o
n
o
f
th
e
K
NN,
R
F a
n
d
Hy
b
r
id
with
R
SC
V
M
e
t
r
i
c
K
N
N
_
R
S
C
V
TP (
7
3
1
7
)
,
TN
(
2
2
1
7
5
)
,
F
P
(
1
1
)
,
F
N
(
2
3
1
)
R
F
_
R
S
C
V
TP (
7
3
2
8
)
,
TN
(
2
2
3
9
4
)
,
F
P
(
1
2
)
,
F
N
(
0
)
H
y
b
r
i
d
_
R
S
C
V
TP (
7
3
2
3
)
,
TN
(
2
2
2
8
8
)
,
F
P
(
1
1
8
)
,
F
N
(
5
)
A
c
c
u
r
a
c
y
:
+
+
+
+
9
9
.
2
%
1
0
0
%
9
9
.
6
%
P
r
e
c
i
s
i
o
n
:
+
9
9
.
8
%
9
9
.
8
%
9
8
.
4
%
R
e
c
a
l
l
:
+
9
6
.
9
%
1
0
0
%
9
9
.
9
%
F1
-
S
c
o
r
e
:
9
8
.
4
%
9
9
.
9
%
9
9
.
2
%
R
O
C
-
A
U
C
:
×
×
+
0
.
9
9
7
1
.
0
0
0
1
.
0
0
0
Ti
me
2
0
.
8
6
7
2
0
.
9
8
6
4
2
1
.
9
3
1
4
Ad
d
itio
n
ally
,
th
e
AUC
-
R
O
C
cu
r
v
e
o
f
th
e
m
o
d
els
with
G
C
SV
an
d
R
S
C
V
p
r
esen
ted
in
Fig
u
r
e
1
6
r
ev
ea
ls
th
at
th
e
GSC
V
an
d
RS
C
V
ac
h
iev
ed
th
e
s
am
e
r
esu
lts
ac
r
o
s
s
th
e
m
etr
ics.
Evaluation Warning : The document was created with Spire.PDF for Python.