I
AE
S
I
nte
rna
t
io
na
l J
o
urna
l o
f
Art
if
icia
l In
t
ellig
ence
(
I
J
-
AI
)
Vo
l.
8
,
No
.
2
,
J
u
n
e
201
9
,
p
p
.
1
68
~
1
7
4
I
SS
N:
2252
-
8938
,
DOI
: 1
0
.
1
1
5
9
1
/i
j
ai.
v
8
.i
2
.
p
p
1
68
-
1
7
4
168
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ia
e
s
co
r
e
.
co
m/o
n
lin
e/in
d
ex
.
p
h
p
/I
J
A
I
Predic
tion o
f
ba
n
k
r
uptcy usi
ng
big da
ta a
na
ly
tic
ba
sed o
n f
u
zz
y
c
-
m
ea
ns a
lg
o
rith
m
Arup
G
uh
a
,
N.
Vee
ra
nja
ney
ulu
V
ig
n
a
n
’s
F
o
u
n
d
a
ti
o
n
f
o
r
S
c
ien
c
e
,
T
e
c
h
n
o
lo
g
y
a
n
d
Re
se
a
rc
h
,
V
a
d
la
m
u
d
i,
In
d
ia
Art
icle
I
nfo
AB
ST
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
F
eb
18
,
2
0
1
9
R
ev
i
s
ed
A
p
r
14
,
2
0
1
9
A
cc
ep
ted
M
ay
16
,
2
0
1
9
T
h
is
p
a
p
e
r
h
a
s
su
g
g
e
ste
d
a
n
o
p
ti
m
iza
ti
o
n
a
p
p
ro
a
c
h
o
f
th
e
c
lu
ste
r
-
b
a
se
d
sa
m
p
li
n
g
u
sin
g
F
u
z
z
y
c
m
e
a
n
s
a
lg
o
rit
h
m
to
th
e
c
las
si
f
ier
in
o
rd
e
r
to
se
lec
t
th
e
m
o
st
a
p
p
r
o
p
riate
in
sta
n
c
e
s
o
f
b
a
n
k
ru
p
tcy
.
T
h
is
m
e
th
o
d
w
a
s
e
x
a
m
in
e
d
w
it
h
th
e
h
e
lp
o
f
a
c
lu
ste
rin
g
m
e
th
o
d
a
n
d
G
A
b
a
se
d
a
rti
f
ici
a
l
n
e
u
ra
l
n
e
tw
o
rk
in
o
r
d
e
r
t
o
so
lv
e
th
e
e
x
isti
n
g
d
a
ta
im
b
a
lan
c
e
issu
e
.
T
h
e
o
b
jec
ti
v
e
o
f
th
is
p
a
p
e
r
is t
o
o
p
ti
m
ize
th
e
se
le
c
ted
d
e
sig
n
m
o
d
e
l
o
f
GA
-
A
N
N b
y
u
sin
g
F
u
z
z
y
C
m
e
a
n
s
a
lg
o
rit
h
m
to
p
re
d
ict
c
o
rp
o
ra
te
b
a
n
k
ru
p
tcie
s
b
y
c
o
n
sid
e
rin
g
d
iff
e
re
n
t
f
in
a
n
c
ial
ra
ti
o
s
o
f
c
o
m
p
a
n
ies
a
c
r
o
ss
se
v
e
ra
l
in
d
u
stries
w
it
h
in
th
e
p
e
rio
d
f
ro
m
1
9
9
4
t
o
2
0
1
4
.
Ef
f
e
c
ti
v
e
n
e
ss
o
f
th
is
m
e
th
o
d
w
a
s
p
ro
v
e
d
b
y
c
o
m
p
a
rin
g
it
s
a
c
c
u
ra
c
y
ra
te
w
it
h
th
e
re
su
lt
s
o
f
e
x
isti
n
g
m
e
th
o
d
.
F
r
o
m
th
e
p
e
rf
o
r
m
a
n
c
e
re
su
lt
th
e
a
c
c
u
ra
c
y
ra
te
o
f
th
is
m
e
th
o
d
w
a
s
f
o
u
n
d
t
o
b
e
7
8
.
2
%
a
n
d
m
is
c
las
si
f
ica
ti
o
n
ra
te t
o
b
e
0
.
2
1
7
8
.
K
ey
w
o
r
d
s
:
A
r
ti
f
icial
n
e
u
r
al
n
et
w
o
r
k
C
lu
s
ter
-
b
a
s
ed
s
a
m
p
li
n
g
Fu
zz
y
c
m
ea
n
s
cl
u
s
ter
i
n
g
Gen
etic
al
g
o
r
ith
m
Ma
ch
i
n
e
lear
n
i
n
g
Un
d
er
-
s
a
m
p
li
n
g
tech
n
iq
u
e
Co
p
y
rig
h
t
©
2
0
1
9
In
stit
u
te o
f
A
d
v
a
n
c
e
d
E
n
g
i
n
e
e
rin
g
a
n
d
S
c
ien
c
e
.
Al
l
rig
h
ts re
se
rv
e
d
.
C
o
r
r
e
s
p
o
nd
ing
A
uth
o
r
:
A
r
u
p
Gu
h
a
,
Vig
n
an
’
s
f
o
u
n
d
atio
n
f
o
r
Scien
ce
,
T
ec
h
n
o
lo
g
y
a
n
d
R
esear
ch
,
Vad
la
m
u
d
i,
I
n
d
ia.
E
m
ail:
g
u
h
a.
ar
u
p
@
g
m
ail.
co
m
1.
I
NT
RO
D
UCT
I
O
N
I
n
cr
ea
s
in
g
a
m
o
u
n
t
o
f
d
ata
h
as
led
to
t
h
e
ev
o
l
u
tio
n
o
f
d
ata
s
cie
n
ce
a
n
d
it
s
ap
p
licatio
n
to
s
o
lv
e
co
m
p
le
x
cla
s
s
i
f
icat
io
n
i
s
s
u
es
in
a
s
et
o
f
d
ata
to
ta
k
e
i
m
p
o
r
tan
t
m
a
n
ag
er
ial
d
ec
i
s
io
n
s
[
1
]
.
T
h
is
p
ap
er
is
b
ased
o
n
an
o
p
ti
m
izatio
n
ap
p
r
o
ac
h
in
A
NN
(
ar
ti
f
icial
n
e
u
r
al
n
et
wo
r
k
)
u
s
i
n
g
t
h
e
co
n
ce
p
t
o
f
Fu
z
z
y
cl
u
s
ter
i
n
g
w
h
ic
h
is
a
t
y
p
e
o
f
s
a
m
p
l
in
g
tech
n
iq
u
e
f
o
r
ex
tr
ac
ti
n
g
ap
p
r
o
p
r
i
ate
in
f
o
r
m
atio
n
f
r
o
m
t
h
e
r
an
d
o
m
d
ata
s
et
[
2
]
.
C
lu
s
ter
i
n
g
i
s
o
n
e
o
f
th
e
e
f
f
e
ctiv
e
f
o
r
m
o
f
d
ata
m
i
n
in
g
t
ec
h
n
iq
u
es
t
h
at
ar
e
w
id
el
y
u
s
ed
f
o
r
p
er
f
o
r
m
in
g
d
escr
ip
tiv
e
lear
n
i
n
g
tech
n
iq
u
e
in
an
al
y
tic
s
f
o
r
p
r
ed
ictin
g
th
e
co
r
p
o
r
ate
b
an
k
r
u
p
tc
y
[
3
]
.
T
h
is
tech
n
iq
u
e
i
s
b
ased
o
n
t
h
e
d
eter
m
i
n
atio
n
o
f
s
i
m
ilar
g
r
o
u
p
s
w
it
h
id
en
tic
al
f
ea
t
u
r
es
a
m
o
n
g
a
h
u
g
e
r
a
n
d
o
m
d
ata
s
et.
T
h
is
m
et
h
o
d
is
a
p
o
p
u
lar
l
y
u
s
ed
s
a
m
p
lin
g
tec
h
n
iq
u
e
i
n
t
h
e
ca
s
e
o
f
i
m
b
ala
n
ce
d
ata
w
it
h
i
n
t
h
e
s
et
o
f
r
a
n
d
o
m
d
ata
b
ec
au
s
e
it
is
v
er
y
d
i
f
f
icu
lt
to
i
d
en
tify
p
atter
n
s
a
m
o
n
g
d
ata
c
o
m
p
r
i
s
in
g
o
f
o
d
d
d
ata
v
alu
es,
eith
er
v
er
y
h
i
g
h
o
r
v
er
y
lo
w
[
4
]
.
T
h
e
m
eth
o
d
o
f
h
an
d
li
n
g
s
u
ch
i
m
b
alan
ce
s
et
o
f
d
ata
is
i
m
p
o
r
tan
t
p
r
io
r
to
m
o
d
el
d
ev
e
lo
p
m
en
t
b
ec
a
u
s
e
i
f
t
h
e
d
if
f
er
en
ce
in
d
ata
s
ize
is
to
o
lar
g
e
o
r
to
o
s
m
all,
t
h
e
n
th
e
ca
s
es
o
f
b
an
k
r
u
p
tc
y
ar
e
i
g
n
o
r
ed
in
th
e
an
al
y
s
is
.
T
h
e
b
asic
m
e
th
o
d
s
th
at
w
er
e
co
n
s
id
er
ed
in
th
i
s
b
an
k
r
u
p
tc
y
p
r
ed
ictio
n
w
er
e
b
ased
o
n
ap
p
ly
in
g
u
n
d
er
s
a
m
p
li
n
g
tec
h
n
iq
u
e
to
th
e
m
aj
o
r
ity
g
r
o
u
p
an
d
o
v
er
-
s
a
m
p
lin
g
to
th
e
m
i
n
o
r
it
y
cla
s
s
.
T
h
e
p
ap
er
is
b
ased
o
n
ap
p
licatio
n
o
f
Gen
et
ic
al
g
o
r
ith
m
(
GA
)
a
n
d
it
s
co
m
b
i
n
atio
n
w
it
h
A
r
ti
f
icial
Neu
r
al
Net
w
o
r
k
(
A
NN)
i.e
G
A
-
ANN
m
o
d
elli
n
g
tech
n
i
q
u
e
[
5
]
.
C
o
n
d
u
cti
n
g
cla
s
s
i
f
i
ca
tio
n
tas
k
s
u
s
i
n
g
u
n
b
ala
n
ce
d
d
ata
u
s
u
al
l
y
d
eter
i
o
r
ates
th
e
class
i
f
ica
tio
n
p
er
f
o
r
m
an
ce
.
I
f
th
e
d
if
f
er
e
n
ce
o
f
th
e
d
ata
s
ize
b
et
w
ee
n
th
e
t
w
o
ca
te
g
o
r
ies
is
g
r
ea
ter
,
m
o
s
t
o
f
th
e
d
ata
is
s
tr
o
n
g
l
y
class
i
f
ied
as
th
e
m
aj
o
r
it
y
cl
ass
to
d
ec
r
ea
s
e
th
e
o
v
er
all
m
is
cla
s
s
i
f
ica
tio
n
[
4
]
.
T
h
er
ef
o
r
e,
h
an
d
lin
g
u
n
b
a
lan
ce
d
d
ata
m
a
y
b
e
a
cr
u
cial
p
r
o
ce
d
u
r
e
in
m
o
d
el
d
ev
elo
p
m
en
t.
T
h
is
r
e
m
ain
s
t
o
b
e
a
m
aj
o
r
d
r
a
w
b
ac
k
i
n
th
e
class
i
f
icatio
n
/p
r
ed
ictio
n
tec
h
n
iq
u
es.
T
h
e
ab
o
v
e
d
r
a
w
b
ac
k
w
i
ll
d
ef
i
n
ite
l
y
h
a
v
e
an
i
m
p
ac
t
o
n
cla
s
s
i
f
ica
tio
n
p
er
f
o
r
m
an
ce
.
T
h
e
p
er
f
o
r
m
a
n
ce
m
etr
ics,
lik
e
th
e
A
U
R
O
C
,
th
e
AR
,
o
r
th
e
H
-
m
e
asu
r
e
h
ad
n
o
d
ef
i
n
ite
cr
iter
ia
t
o
p
r
o
d
u
ce
ev
id
en
ce
f
o
r
ev
al
u
a
tin
g
t
h
e
ex
ce
l
len
c
e
Evaluation Warning : The document was created with Spire.PDF for Python.
IJ
-
AI
I
SS
N:
2252
-
8938
P
r
ed
ictio
n
o
f b
a
n
kru
p
tcy
u
s
in
g
b
ig
d
a
ta
a
n
a
lytic
b
a
s
ed
o
n
f
u
z
z
y
c
-
mea
n
s
a
lg
o
r
ith
m
.
.
.
(
A
r
u
p
Gu
h
a
)
169
o
f
th
e
m
o
d
el
p
er
f
o
r
m
an
ce
.
R
e
f
i
n
in
g
th
e
d
ata
w
ill
ad
d
to
th
e
p
er
f
o
r
m
a
n
ce
i
m
p
r
o
v
e
m
e
n
t
o
f
A
N
N
as
it
ca
n
k
ee
p
a
ch
ec
k
o
n
t
h
e
co
m
p
u
ta
tio
n
ti
m
e
a
n
d
al
s
o
r
ed
u
ce
s
p
en
d
i
n
g
ex
tr
a
co
m
p
u
ti
n
g
r
eso
u
r
ce
s
o
n
tr
ain
i
n
g
A
NN
s
[
5
]
.
Op
ti
m
izi
n
g
th
e
r
eq
u
ir
ed
d
ata
w
il
l
aid
in
p
r
o
v
id
in
g
i
m
p
r
o
v
e
d
class
if
icatio
n
ac
c
u
r
ac
y
a
n
d
th
er
eb
y
en
h
a
n
ci
n
g
th
e
r
esu
lts
o
f
p
r
ed
ictio
n
.
T
h
e
p
r
o
p
o
s
ed
m
eth
o
d
is
ap
p
li
ed
to
th
e
p
r
o
b
lem
o
f
b
an
k
r
u
p
tc
y
p
r
ed
ictio
n
s
u
s
i
n
g
t
h
e
f
i
n
a
n
cial
d
ata
th
at
w
er
e
co
llected
in
o
r
d
er
t
o
f
o
cu
s
o
n
t
h
e
p
r
o
p
o
r
tio
n
o
f
th
e
s
m
all
an
d
m
ed
iu
m
-
s
ca
le
b
an
k
r
u
p
tc
y
f
ir
m
s
.
T
h
e
in
te
n
tio
n
o
f
th
e
s
tu
d
y
i
s
to
clar
if
y
a
n
d
i
n
v
e
s
ti
g
a
te
h
o
w
a
m
ac
h
in
e
lear
n
in
g
tech
n
iq
u
e
ca
n
b
e
e
x
p
lo
ited
w
it
h
i
n
t
h
e
f
ie
ld
o
f
ec
o
n
o
m
ic
s
.
Mo
r
e
s
p
ec
if
icall
y
,
ai
m
o
f
t
h
e
r
esear
ch
is
to
r
ef
i
n
e
h
o
w
t
h
e
m
ac
h
i
n
e
lear
n
in
g
s
tr
ateg
ie
s
co
u
ld
b
e
h
ar
n
ess
ed
to
p
r
ed
ict
co
r
p
o
r
ate
b
an
k
r
u
p
tc
ies.
W
e
in
ten
t
to
ap
p
l
y
a
n
ap
p
r
o
ac
h
f
o
r
s
elec
tin
g
th
e
o
p
ti
m
al
tr
ain
in
g
d
ata
s
et
an
d
f
o
u
n
d
a
p
r
o
p
er
c
o
n
n
ec
tio
n
w
e
ig
h
t
to
lear
n
th
e
A
NN
m
o
d
el
w
h
er
e
w
e
ca
n
e
m
p
lo
y
m
u
lti
-
m
o
d
al
G
A
u
s
i
n
g
F
u
zz
y
C
m
ea
n
s
alg
o
r
it
h
m
t
o
f
in
d
m
u
l
tip
le
s
o
lu
tio
n
s
o
n
t
h
e
cu
t
-
o
f
f
v
al
u
es
o
f
ev
er
y
cl
u
s
ter
.
T
h
is
w
a
y
b
y
e
m
p
lo
y
i
n
g
clu
s
ter
i
n
g
an
d
o
p
ti
m
al
s
elec
tio
n
ap
p
r
o
ac
h
th
e
n
e
u
r
al
n
e
t
w
o
r
k
s
w
ill
b
e
m
o
r
e
i
m
p
r
o
v
ed
b
ec
au
s
e
th
e
f
ea
tu
r
e
s
elec
tio
n
m
et
h
o
d
to
id
e
n
ti
f
y
t
h
e
m
o
s
t
ef
f
ec
ti
v
e
f
ea
t
u
r
es
f
o
r
th
e
class
if
ier
w
il
l
en
h
an
ce
t
h
e
ac
cu
r
ac
y
o
f
t
h
eir
p
r
ed
ictio
n
o
f
co
r
p
o
r
ate
b
a
n
k
r
u
p
tc
y
[
4
]
.
T
h
e
r
em
ai
n
in
g
s
ec
tio
n
o
f
th
e
p
ap
er
is
o
r
g
a
n
ized
i
n
th
e
f
o
llo
w
i
n
g
w
a
y
.
Sectio
n
2
d
escr
ib
es
t
h
e
e
x
is
t
in
g
tech
n
iq
u
e
o
f
s
a
m
p
li
n
g
an
d
h
o
w
s
a
m
p
l
in
g
tech
n
iq
u
e
h
a
s
b
ee
n
u
s
ed
b
y
m
an
y
r
esear
ch
er
s
o
v
er
th
e
p
er
io
d
o
f
tim
e
i
n
o
r
d
er
to
s
o
lv
e
co
m
p
le
x
is
s
u
e
s
o
f
i
m
b
alan
c
e
d
ata
m
a
n
ag
e
m
e
n
t.
Sectio
n
3
m
en
tio
n
ed
t
h
e
p
r
o
p
o
s
ed
Fu
zz
y
clu
s
ter
-
b
ased
t
ec
h
n
iq
u
e
o
f
s
o
lv
i
n
g
b
an
k
r
u
p
tc
y
p
r
o
b
le
m
i
n
a
m
o
r
e
s
p
ec
if
ic
m
a
n
n
er
.
Sec
tio
n
4
h
as
p
r
esen
ted
t
h
e
o
u
tco
m
e
o
f
t
h
e
p
r
o
p
o
s
ed
tech
n
iq
u
e
i
n
th
e
f
o
r
m
o
f
th
e
ir
i
m
p
le
m
en
ta
tio
n
an
d
e
x
p
er
i
m
e
n
tal
r
e
s
u
lts
.
Sectio
n
5
b
r
ief
s
ab
o
u
t
th
e
co
n
cl
u
s
io
n
o
f
th
e
r
es
u
lt c
o
n
s
id
er
in
g
th
e
f
i
n
d
in
g
s
o
f
th
e
p
r
o
p
o
s
ed
alg
o
r
ith
m
w
ith
r
e
s
ea
r
ch
g
ap
s
a
n
d
li
m
itatio
n
s
.
2.
L
I
T
E
R
AT
U
RE
R
E
VI
E
W
P
r
ev
io
u
s
s
t
u
d
ies
i
n
s
o
l
v
in
g
d
ata
i
m
b
alan
ce
p
r
o
b
le
m
w
er
e
r
ef
er
r
ed
at
th
e
t
w
o
ap
p
r
o
ac
h
es
o
f
d
ata
lev
el
an
d
alg
o
r
it
h
m
le
v
el.
T
h
e
v
ar
io
u
s
co
n
ce
p
ts
,
tech
n
iq
u
e
s
an
d
s
y
s
te
m
s
ar
e
d
is
cu
s
s
ed
i
n
th
i
s
s
ec
tio
n
b
ased
o
n
th
e
ex
is
ti
n
g
r
esear
ch
i
n
th
e
cu
r
r
en
t sce
n
ar
io
.
2
.
1
.
Undersa
m
p
lin
g
t
ec
hn
i
q
ue
Un
d
er
s
a
m
p
li
n
g
tech
n
iq
u
e
i
s
r
ef
er
r
ed
to
class
if
ica
tio
n
i
n
ter
m
s
o
f
r
ed
u
ctio
n
i
n
th
e
n
u
m
b
er
o
f
in
s
ta
n
ce
s
to
b
alan
ce
t
h
e
d
atas
et
co
n
s
i
s
ti
n
g
o
f
m
aj
o
r
ity
cla
s
s
an
d
th
e
m
i
n
o
r
it
y
c
lass
.
T
h
is
i
s
an
e
f
f
icie
n
t
m
o
d
el
in
t
h
e
ca
s
e
o
f
d
ea
lin
g
w
i
th
l
ar
g
e
a
m
o
u
n
t
o
f
d
ata.
T
h
is
te
ch
n
iq
u
e
is
h
elp
f
u
l
s
i
n
ce
t
h
e
t
r
ain
in
g
ti
m
e
o
f
t
h
e
d
ataset
is
r
ed
u
ce
d
.
Ho
w
e
v
er
,
t
h
is
m
et
h
o
d
p
o
s
s
es
s
d
is
ad
v
a
n
t
ag
es
i
n
t
h
e
f
o
r
m
o
f
r
is
k
o
f
d
is
to
r
tin
g
th
e
o
r
i
g
i
n
al
d
is
tr
ib
u
tio
n
o
f
t
h
e
m
aj
o
r
it
y
c
lass
.
Mo
r
eo
v
er
,
in
t
h
i
s
tech
n
i
q
u
e
th
e
p
o
ten
tial
u
s
e
f
u
l
d
ata
is
d
is
ca
r
d
ed
.
I
t
is
cr
u
cial
to
h
a
v
e
a
r
ele
v
a
n
t
d
ata
s
et
to
i
m
p
r
o
v
e
t
h
e
clas
s
if
icati
o
n
p
er
f
o
r
m
a
n
ce
o
f
a
m
o
d
el
b
y
s
a
m
p
l
in
g
d
ata
w
it
h
s
i
m
ilar
p
r
o
p
er
ties
.
R
an
d
o
m
u
n
d
er
s
a
m
p
li
n
g
r
ed
u
ce
s
t
h
e
d
ataset
b
y
r
e
m
o
v
i
n
g
a
r
an
d
o
m
l
y
s
a
m
p
led
d
ataset
f
r
o
m
t
h
e
m
aj
o
r
it
y
class
as
t
h
e
s
i
m
p
les
t
m
et
h
o
d
.
Ho
w
ev
er
,
p
ar
tial
d
ata
ca
n
also
b
e
u
s
ed
in
d
ata
m
o
d
eli
n
g
b
ec
au
s
e
th
is
h
u
g
e
a
m
o
u
n
t o
f
d
ata
is
s
u
f
f
icie
n
t
f
o
r
an
al
y
s
is
i
n
th
e
er
a
o
f
b
ig
d
ata
[
4
].
A
cl
u
s
ter
-
b
ased
u
n
d
er
s
a
m
p
li
n
g
ap
p
r
o
ac
h
w
as
p
er
f
o
r
m
ed
w
h
er
e
th
e
ap
p
r
o
ac
h
h
as
f
ir
s
t
co
n
d
u
cted
clu
s
ter
i
n
g
o
f
a
ll
i
n
s
ta
n
ce
s
o
f
d
ata
an
d
d
iv
id
ed
th
e
m
in
to
s
ev
er
al
cl
u
s
ter
s
[
6
]
.
Ne
x
t,
it
s
elec
ts
t
h
e
p
o
ten
tial
r
elev
an
t
n
u
m
b
er
o
f
in
s
ta
n
ce
s
th
at
i
s
b
elo
n
g
in
g
to
t
h
e
m
a
j
o
r
ity
clas
s
f
r
o
m
ea
ch
c
lu
s
te
r
o
n
th
e
b
asi
s
o
f
p
r
o
p
o
r
tio
n
al
in
s
ta
n
ce
s
m
aj
o
r
it
y
clas
s
to
t
h
e
n
u
m
b
er
o
f
i
n
s
tan
ce
s
o
f
t
h
e
m
i
n
o
r
it
y
cla
s
s
w
i
th
i
n
t
h
e
cl
u
s
ter
.
C
lu
s
ter
i
n
g
,
en
s
e
m
b
le
a
n
d
u
n
d
er
s
a
m
p
li
n
g
m
et
h
o
d
s
w
er
e
p
er
f
o
r
m
ed
in
o
n
e
s
t
u
d
y
to
s
o
l
v
e
th
e
class
i
m
b
ala
n
ce
p
r
o
b
lem
[
7
]
.
T
h
ey
f
ir
s
t
co
n
d
u
cted
clu
s
ter
i
n
g
u
s
in
g
in
s
tan
ce
s
o
f
th
e
m
aj
o
r
it
y
class
a
n
d
th
en
co
n
s
tr
u
cted
m
u
ltip
le
tr
ai
n
in
g
d
ataset
s
co
m
p
r
is
in
g
o
f
s
a
m
p
led
i
n
s
ta
n
ce
s
o
f
t
h
e
m
aj
o
r
it
y
cla
s
s
f
r
o
m
ea
c
h
clu
s
ter
,
p
r
eser
v
i
n
g
in
s
ta
n
ce
s
o
f
t
h
e
m
i
n
o
r
it
y
cla
s
s
.
T
h
e
ev
o
lu
tio
n
ar
y
s
a
m
p
li
n
g
m
et
h
o
d
b
ased
o
n
G
A
h
a
s
b
ee
n
d
ep
lo
y
ed
in
o
r
d
er
to
s
elec
ti
v
el
y
r
e
m
o
v
e
i
n
s
tan
c
es
f
r
o
m
t
h
e
m
aj
o
r
it
y
cla
s
s
[
8
,
9
]
.
Ho
w
ev
er
,
p
r
ev
io
u
s
s
t
u
d
i
es
o
n
ev
o
l
u
tio
n
ar
y
s
a
m
p
li
n
g
u
s
in
g
G
A
h
av
e
s
h
o
w
ed
p
er
f
o
r
m
a
n
ce
r
esu
lts
o
f
ti
m
e
-
co
n
s
u
m
i
n
g
ta
s
k
s
in
e
x
p
lo
r
in
g
o
p
ti
m
al
o
r
n
ea
r
o
p
tim
a
l
s
o
l
u
tio
n
s
,
s
in
ce
i
n
s
ta
n
ce
s
o
f
t
h
e
m
aj
o
r
it
y
cla
s
s
h
a
s
b
ec
o
m
e
s
t
r
i
n
g
s
f
o
r
G
A
s
ea
r
c
h
in
g
.
T
h
u
s
,
i
n
th
i
s
s
tu
d
y
a
cl
u
s
ter
-
b
ased
s
a
m
p
l
in
g
s
u
p
p
o
r
ted
b
y
G
A
is
s
u
g
g
est
ed
in
o
r
d
er
to
h
a
n
d
le
t
h
e
i
n
-
e
f
f
icie
n
c
y
p
r
o
b
lem
o
f
th
e
p
r
ev
io
u
s
ex
i
s
ti
n
g
e
v
o
lu
tio
n
ar
y
s
a
m
p
li
n
g
m
et
h
o
d
.
2
.
2
.
Clus
t
er
ing
o
f
no
n
-
ba
nk
ruptc
y
f
ir
m
da
t
a
ba
s
ed
o
n
ma
j
o
rit
y
cla
s
s
A
cl
u
s
ter
b
ased
b
o
o
s
tin
g
al
g
o
r
ith
m
w
as
p
er
f
o
r
m
ed
in
o
n
e
s
t
u
d
y
u
s
i
n
g
th
e
I
n
s
ta
n
ce
Har
d
n
ess
T
h
r
esh
o
ld
an
d
C
B
o
o
s
t
alg
o
r
ith
m
w
i
th
a
r
o
b
u
s
t
f
r
a
m
e
w
o
r
k
in
o
r
d
er
to
p
r
ed
ict
b
an
k
r
u
p
tcy
e
f
f
ec
ti
v
el
y
o
f
th
e
f
i
n
an
cia
l
i
m
b
ala
n
ce
d
ataset
[
3
]
.
T
h
is
p
r
o
p
o
s
ed
f
r
a
m
e
w
o
r
k
i
s
also
v
er
if
ied
b
y
t
h
e
KB
D
(
Ko
r
ea
n
b
an
k
r
u
p
tc
y
d
ataset)
h
a
v
i
n
g
a
s
m
all
b
alan
cin
g
r
atio
in
b
o
t
h
t
h
e
te
s
ti
n
g
an
d
tr
ai
n
i
n
g
p
h
ase
s
.
T
h
e
p
r
o
p
o
s
ed
m
o
d
el
ex
p
er
i
m
e
n
t
r
es
u
lt
s
h
a
s
ac
h
iev
ed
8
6
.
8
%
in
A
U
C
i.e
.
t
h
e
ar
ea
u
n
d
er
R
OC
c
u
r
v
e.
I
t
h
a
s
also
o
u
tp
er
f
o
r
m
ed
o
th
er
ex
is
t
in
g
m
e
th
o
d
s
f
o
r
b
an
k
r
u
p
tc
y
p
r
ed
ictio
n
u
s
i
n
g
i
m
b
ala
n
ce
s
et
o
f
d
ata.
Ma
ch
in
e
lear
n
in
g
m
et
h
o
d
s
w
er
e
ap
p
lied
to
th
e
d
ataset
co
llected
f
r
o
m
t
h
e
m
a
n
u
f
ac
tu
r
i
n
g
co
m
p
a
n
ie
s
i
n
Ko
r
ea
,
in
o
r
d
er
to
k
n
o
w
t
h
eir
f
u
tu
r
e
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2252
-
8
9
3
8
IJ
-
AI
Vo
l.
8
,
No
.
2
,
J
u
n
e
20
1
9
:
1
68
–
1
7
4
170
s
tate
w
it
h
t
h
e
h
elp
o
f
ce
r
tain
f
in
an
c
ial
m
ea
s
u
r
es
[
10
]
.
Usi
n
g
s
ev
er
al
m
ac
h
i
n
e
lear
n
in
g
m
et
h
o
d
r
esu
lt
s
h
o
w
ed
an
ac
c
u
r
ac
y
o
f
m
o
r
e
t
h
a
n
9
5
%.
Ho
w
ev
er
,
th
i
s
s
t
u
d
y
h
as
s
o
m
e
li
m
ita
tio
n
a
ls
o
i
n
th
e
f
o
r
m
o
f
d
i
m
en
s
io
n
a
l
is
s
u
e.
2
.
3
.
Under
s
a
m
pl
ing
t
ec
hn
iq
ue
ba
s
ed
o
n g
enet
ic
a
lg
o
rit
hm
s
(
G
A
)
A
r
e
-
s
a
m
p
li
n
g
ap
p
r
o
ac
h
is
p
er
f
o
r
m
ed
i
n
a
s
t
u
d
y
i
n
o
r
d
er
to
s
o
lv
e
t
h
e
u
n
b
ala
n
ce
d
d
ata
s
et
s
[
5
]
.
I
n
th
i
s
ap
p
r
o
ac
h
,
b
o
th
th
e
o
v
er
s
a
m
p
lin
g
an
d
u
n
d
er
s
a
m
p
li
n
g
m
et
h
o
d
ar
e
co
m
b
in
ed
w
it
h
th
e
h
elp
o
f
g
e
n
etic
alg
o
r
ith
m
(
G
A
)
.
T
h
e
ap
p
licatio
n
o
f
g
en
etic
a
lg
o
r
it
h
m
i
s
b
ased
o
n
a
s
et
o
f
d
eter
m
i
n
ed
cr
iter
ia
an
d
th
e
u
n
b
alan
ce
r
ate.
T
h
is
ap
p
r
o
ac
h
h
a
s
b
ee
n
te
s
ted
o
n
liter
atu
r
e
as
w
ell
as
in
d
u
s
tr
ial
d
at
asets
a
n
d
a
d
esire
d
i
m
p
r
o
v
e
m
en
t o
n
th
e
cla
s
s
i
f
ica
tio
n
p
er
f
o
r
m
a
n
ce
h
as b
ee
n
o
b
s
er
v
ed
[
11
].
An
u
n
d
er
s
a
m
p
li
n
g
ap
p
r
o
ac
h
an
d
GA
-
A
N
N
m
o
d
el
h
a
s
b
ee
n
ap
p
r
o
ac
h
ed
in
a
s
tu
d
y
t
o
im
p
r
o
v
e
th
e
ex
is
ti
n
g
tr
ad
itio
n
al
ap
p
r
o
ac
h
o
f
clas
s
i
f
icatio
n
,
w
h
ic
h
w
er
e
u
s
u
all
y
co
s
tl
y
a
n
d
s
lo
w
[
5
]
.
T
h
e
u
n
d
er
s
a
m
p
li
n
g
ap
p
r
o
ac
h
is
b
ased
o
n
K
-
m
ea
n
s
clu
s
ter
d
is
tr
ib
u
t
io
n
i
n
o
r
d
e
r
to
s
o
lv
e
t
h
e
p
r
o
b
lem
s
o
f
i
m
b
alan
ce
s
et
o
f
d
ata.
T
h
is
m
e
th
o
d
i
s
e
f
f
ec
tiv
e
to
e
n
h
a
n
ce
th
e
r
ate
o
f
s
a
m
p
li
n
g
a
n
d
i
m
p
r
o
v
ed
th
e
f
in
a
l
cla
s
s
i
f
i
ca
tio
n
.
A
t
th
e
s
a
m
e
ti
m
e,
t
h
i
s
m
et
h
o
d
h
as
lo
w
er
ti
m
e
o
f
p
r
o
ce
s
s
in
g
.
G
A
-
ANN
m
et
h
o
d
u
s
ed
i
n
t
h
eir
s
t
u
d
y
u
s
es
th
e
al
g
o
r
it
h
m
to
o
p
tim
ize
t
h
e
b
ias
a
n
d
w
eig
h
t
o
f
th
e
n
e
u
r
al
n
e
t
w
o
r
k
a
n
d
th
er
eb
y
r
es
u
lted
i
n
to
b
etter
p
er
f
o
r
m
a
n
ce
.
T
o
in
cr
ea
s
i
n
g
t
h
e
class
if
icatio
n
a
cc
u
r
ac
y
a
n
e
w
g
e
n
etic
al
g
o
r
ith
m
w
a
s
p
r
o
p
o
s
ed
b
ased
o
n
o
v
er
s
a
m
p
lin
g
in
o
r
d
er
to
s
o
lv
e
th
e
cla
s
s
i
m
b
alan
ce
d
ata
s
ets
[
12
]
.
I
t
ca
n
cr
ea
te
o
p
tim
ized
m
i
n
o
r
it
y
clas
s
ev
en
ts
to
b
alan
ce
t
h
e
tr
ain
i
n
g
d
ataset
s
.
T
h
e
ex
p
er
im
en
tal
r
esu
l
ts
o
n
i
m
b
alan
ce
d
d
atasets
p
r
o
v
ed
b
etter
p
er
f
o
r
m
a
n
ce
o
v
er
th
e
p
r
ev
io
u
s
s
a
m
p
li
n
g
m
et
h
o
d
s
in
ter
m
s
o
f
AUC a
n
d
F
-
m
ea
s
u
r
e.
3.
P
RO
P
O
SE
D
M
O
DE
L
T
h
e
p
r
o
p
o
s
ed
clu
s
ter
-
b
ased
m
eth
o
d
is
b
ased
o
n
a
clu
s
ter
in
g
alg
o
r
ith
m
.
I
n
th
i
s
s
tu
d
y
,
th
e
m
et
h
o
d
is
ad
o
p
ted
u
s
in
g
F
u
zz
y
C
cl
u
s
ter
in
g
.
F
u
zz
y
c
-
m
ea
n
s
alg
o
r
i
th
m
ap
p
lies
t
h
e
co
n
ce
p
t
o
f
f
u
zz
y
lo
g
ic
w
h
er
e
th
e
o
b
j
ec
ts
o
f
class
i
f
icat
io
n
s
ar
e
allo
w
ed
f
o
r
m
o
r
e
t
h
an
o
n
e
clu
s
ter
.
T
h
i
s
t
y
p
e
o
f
cla
s
s
i
f
icatio
n
m
ak
e
s
h
ig
h
clar
it
y
s
e
n
s
e
s
i
n
ce
all
th
e
cl
u
s
ter
s
ar
e
w
ell
s
ep
ar
ated
.
I
n
th
i
s
tech
n
iq
u
e,
v
al
u
e
ar
e
ass
i
g
n
e
d
to
all
th
e
w
ig
h
t
s
.
R
ep
etitio
n
is
d
o
n
e
u
n
til
t
h
e
ce
n
tr
o
id
is
co
m
p
u
ted
f
o
r
ea
ch
o
f
th
e
c
lu
s
ter
w
it
h
t
h
e
h
elp
o
f
f
u
zz
y
p
ar
titi
o
n
.
T
h
i
s
co
n
ce
p
t
is
r
elate
d
w
it
h
t
h
e
d
ev
elo
p
m
en
t
o
f
k
-
m
ea
n
s
al
g
o
r
ith
m
f
o
r
t
h
e
s
e
n
s
o
r
n
e
t
w
o
r
k
.
Usi
n
g
t
h
e
f
u
zz
y
c
-
m
ea
n
s
al
g
o
r
ith
m
t
h
e
p
ar
tit
io
n
in
g
o
f
d
ata
i
s
p
o
s
s
ib
le
b
y
t
h
e
n
o
d
es
i
n
to
d
if
f
er
en
t
m
ea
s
u
r
e
-
d
ep
en
d
e
n
t
s
et
o
f
g
r
o
u
p
s
[
1
3
]
.
T
h
e
r
o
le
o
f
th
i
s
alg
o
r
it
h
m
is
to
class
if
y
t
h
e
d
ata
in
to
s
ep
ar
ate
g
r
o
u
p
s
.
E
a
ch
o
f
t
h
e
s
ep
ar
ated
g
r
o
u
p
s
ar
e
th
e
n
u
s
ed
to
f
i
n
d
o
u
t
th
e
ce
n
tr
o
id
s
an
d
b
ased
o
n
th
e
s
e,
h
i
g
h
p
r
io
r
it
y
an
d
lo
w
p
r
io
r
ity
v
alu
e
s
ar
e
d
eter
m
in
ed
f
o
r
th
e
b
an
k
r
u
p
tc
y
an
d
n
o
n
b
an
k
r
u
p
tc
y
d
at
a.
T
h
e
p
u
r
p
o
s
e
o
f
th
is
n
e
w
l
y
p
r
o
p
o
s
ed
m
o
d
el
is
to
d
eter
m
in
e
t
h
e
r
is
k
o
f
b
an
k
r
u
p
tc
y
w
i
th
i
n
t
h
ese
p
r
ed
icted
r
an
g
e
o
f
g
at
h
er
ed
d
ata,
co
n
s
id
er
in
g
1
2
s
et
o
f
attr
ib
u
tes.
I
n
o
u
r
p
r
o
p
o
s
ed
i
m
p
le
m
e
n
tat
io
n
w
e
ar
e
u
s
i
n
g
en
h
a
n
ce
d
GANN
b
ased
m
u
lti
m
o
d
al
G
A
b
ased
n
eu
r
al
n
et
w
o
r
k
.
C
o
n
s
tan
t c
ap
ital o
r
f
i
x
ed
ass
et
s
.
C
u
r
r
en
t a
s
s
et
s
,
in
v
e
n
to
r
y
an
d
r
ec
eiv
ab
les o
r
s
h
o
r
t
-
ter
m
liab
il
i
ties
(
R
ec
eiv
ab
les
*
3
6
5
)
/ to
tal
ass
ets
(
Net
p
r
o
f
it +
d
ep
r
ec
iatio
n
)
/ to
tal
ass
et
s
T
o
tal
s
ales /
to
tal
ass
ets
Sh
o
r
t
-
ter
m
liab
ilit
ies / to
tal
as
s
ets
W
o
r
k
in
g
ca
p
ital /
to
tal
as
s
ets
W
o
r
k
in
g
ca
p
ital /
s
a
les
(
C
u
r
r
en
t liab
il
ities
*
3
6
5
)
/ c
o
s
t o
f
p
r
o
d
u
cts s
o
ld
(
C
u
r
r
en
t a
s
s
et
s
-
i
n
v
en
to
r
y
-
r
e
ce
iv
ab
les)
/ lo
n
g
-
ter
m
liab
il
iti
es
(
I
n
v
e
n
to
r
y
*
3
6
5
)
/ sales
Net
p
r
o
f
it/i
n
v
e
n
to
r
y
T
h
e
s
tep
b
y
s
tep
p
r
o
ce
s
s
o
f
t
h
e
p
r
o
p
o
s
ed
m
o
d
el
is
s
h
o
w
n
i
n
th
e
f
ig
u
r
e.
Fig
u
r
e:
P
r
o
p
o
s
ed
m
o
d
el
o
f
b
an
k
r
u
p
tc
y
p
r
ed
ictio
n
T
h
e
p
r
o
ce
s
s
o
f
th
e
m
o
d
el
co
m
p
r
is
es o
f
th
e
f
o
llo
w
i
n
g
p
r
o
ce
s
s
:
Step
1
:
I
n
th
e
f
ir
s
t
s
tep
o
f
th
e
m
o
d
el
d
esi
g
n
,
w
e
h
a
v
e
g
at
h
er
ed
th
e
f
in
a
n
cial
d
ata
o
f
co
m
p
an
ies
ac
r
o
s
s
s
e
v
er
al
in
d
u
s
tr
ies
in
I
n
d
ia
alo
n
g
w
it
h
th
eir
d
if
f
er
en
t
f
in
a
n
cial
r
atio
s
w
it
h
i
n
th
e
p
er
io
d
1
9
9
4
to
2
0
1
4
.
B
ig
Data
r
elate
d
to
b
an
k
r
u
p
tc
y
is
co
n
s
id
er
ed
.
T
h
ese
s
et
o
f
b
a
n
k
r
u
p
tc
y
an
d
n
o
n
-
b
an
k
r
u
p
tc
y
d
ata
ar
e
b
ein
g
s
to
r
ed
in
m
er
g
ed
d
ata_
1
0
X.
csv
.
T
h
e
d
ata
is
th
en
p
r
ep
r
o
ce
s
s
ed
to
clea
n
n
o
is
e
d
ata,
n
u
ll
d
ata
an
d
m
i
s
s
i
n
g
d
ata
an
d
th
en
s
to
r
ed
in
tr
an
s
f
o
r
m
ed
_
n
e
w
d
ata.
cs
v
b
y
cr
ea
tin
g
a
s
p
ec
if
ic
p
ath
o
f
p
r
ep
r
o
ce
s
s
ed
d
ata.
T
h
e
F
i
g
u
r
e
1
s
h
o
w
s
cl
u
s
ter
ed
d
ata
alo
n
g
w
it
h
t
h
eir
ce
n
tr
o
id
s
,
u
s
i
n
g
Fu
zz
y
c
m
ea
n
s
cl
u
s
ter
in
g
.
Evaluation Warning : The document was created with Spire.PDF for Python.
IJ
-
AI
I
SS
N:
2252
-
8938
P
r
ed
ictio
n
o
f b
a
n
kru
p
tcy
u
s
in
g
b
ig
d
a
ta
a
n
a
lytic
b
a
s
ed
o
n
f
u
z
z
y
c
-
mea
n
s
a
lg
o
r
ith
m
.
.
.
(
A
r
u
p
Gu
h
a
)
171
Fig
u
r
e
1
.
Fu
zz
y
c
-
m
ea
n
s
cl
u
s
t
er
in
g
Step
2
:
Data
g
at
h
er
ed
is
p
r
ep
r
o
ce
s
s
ed
u
s
in
g
u
n
d
er
g
o
n
e
f
u
zz
y
c
m
ea
n
s
al
g
o
r
ith
m
a
n
d
f
o
llo
w
ed
b
y
d
ata
f
ilter
i
n
g
.
W
it
h
th
e
h
e
lp
o
f
th
is
d
ata,
a
1
2
*
1
2
co
r
r
el
atio
n
m
atr
i
x
is
f
o
r
m
ed
co
n
s
id
er
in
g
ea
ch
o
f
th
e
attr
ib
u
tes.
T
h
en
t
h
e
m
atr
i
x
h
a
s
b
ee
n
ar
r
an
g
ed
co
n
s
id
er
in
g
th
eir
co
r
r
elatio
n
h
ea
t
m
ap
.
T
h
e
F
ig
u
r
e
2
s
h
o
w
s
th
e
co
r
r
elatio
n
h
ea
t
m
ap
.
W
ith
th
is
m
atr
i
x
,
m
a
x
i
m
u
m
p
r
io
r
ity
ca
n
b
e
d
eter
m
i
n
ed
o
f
ea
ch
attr
ib
u
te
v
al
u
es
w
it
h
th
e
h
elp
o
f
co
r
r
elatio
n
m
atr
i
x
.
Fig
u
r
e
2
.
C
o
r
r
elatio
n
m
atr
ices
w
it
h
h
ea
t
m
ap
Step
3
:
T
h
ese
s
et
o
f
attr
ib
u
te
clu
s
ter
ed
d
ata
is
th
e
n
an
al
y
s
ed
w
it
h
th
e
h
elp
o
f
h
i
s
to
g
r
a
m
in
o
r
d
er
to
p
r
ed
ic
t
b
an
k
r
u
p
tc
y
a
n
d
n
o
b
an
k
r
u
p
tcy
d
ata,
as
s
h
o
w
n
in
t
h
e
f
i
g
u
r
e
.
Had
o
o
p
m
ap
r
ed
u
ce
alg
o
r
ith
m
h
as
b
ee
n
ap
p
lied
to
th
ese
p
r
ep
r
o
ce
s
s
ed
d
ata.
Step
4
:
B
an
k
r
u
p
tc
y
an
d
n
o
n
-
b
an
k
r
u
p
tc
y
s
ta
tu
s
o
f
d
ata
is
f
o
u
n
d
w
it
h
th
e
f
ir
s
t
attr
ib
u
te
i.e
co
n
s
tan
t
ca
p
ital
o
r
f
i
x
ed
ass
ets.
L
ik
e
w
i
s
e
w
e
h
a
v
e
p
r
o
ce
e
d
ed
w
it
h
ea
ch
attr
ib
u
t
e.
T
h
e
m
atr
ice
s
w
er
e
d
eter
m
i
n
ed
alo
n
g
w
it
h
h
ea
t
m
ap
th
at
ar
e
clas
s
i
f
ied
co
lo
u
r
w
i
s
e
w
it
h
attr
ib
u
te
s
r
an
g
e
as
s
h
o
w
n
i
n
t
h
e
F
ig
u
r
e
3
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2252
-
8
9
3
8
IJ
-
AI
Vo
l.
8
,
No
.
2
,
J
u
n
e
20
1
9
:
1
68
–
1
7
4
172
Fig
u
r
e
3
.
His
to
g
r
a
m
a
n
al
y
s
i
s
o
f
p
r
o
ce
s
s
ed
d
ata
Ag
g
lo
m
er
ati
v
e
h
ier
ar
ch
ical
C
lu
s
ter
tec
h
n
iq
u
e
h
as
b
ee
n
e
m
p
lo
y
ed
i
n
t
h
i
s
ca
s
e
to
i
m
p
r
o
v
e
th
e
e
f
f
icien
c
y
o
f
t
h
e
b
an
k
r
u
p
t
c
y
.
Af
ter
p
er
f
o
r
m
i
n
g
cl
u
s
ter
i
n
g
o
n
th
e
e
x
tr
ac
ted
attr
ib
u
tes,
th
e
clu
s
te
r
fea
tu
r
e
ve
cto
r
is
ap
p
lied
to
m
o
d
if
y
t
h
e
class
i
f
ier
s
f
o
r
p
r
ed
ictin
g
b
an
k
r
u
p
tc
y
f
r
o
m
th
e
d
ata.
Step
5
:
T
h
e
p
r
ep
r
o
ce
s
s
ed
d
ata
an
d
t
h
e
cl
u
s
ter
ed
d
ata
is
s
to
r
ed
in
to
t
h
e
tr
a
n
s
f
o
r
m
ed
_
n
e
w
d
ata.
cs
v
.
T
h
e
f
ile
is
cr
ea
ted
au
to
m
atic
all
y
an
d
r
en
a
m
ed
as
d
ata.
csv
,
w
h
ich
i
s
o
u
r
m
ai
n
d
ata.
T
h
is
m
ai
n
d
ata
is
n
o
w
s
ep
ar
ated
in
to
test
i
n
g
d
ata
an
d
tr
ain
i
n
g
d
ata
f
o
r
th
e
p
r
ed
ictio
n
o
f
b
a
n
k
r
u
p
tc
y
b
y
co
n
s
id
er
in
g
t
h
e
m
w
it
h
t
h
e
s
et
o
f
1
2
attr
ib
u
te.
T
h
e
class
i
f
ic
atio
n
is
d
o
n
e
w
it
h
clas
s
i
f
ier
s
u
p
p
o
r
t
v
ec
to
r
m
ac
h
i
n
e,
lo
g
i
s
tic
r
eg
r
es
s
io
n
a
n
d
GA
-
A
N
N
in
o
r
d
er
to
co
m
p
ar
e.
Step
6
:
B
ef
o
r
e
class
if
ica
tio
n
o
f
th
e
d
ata
is
d
o
n
e,
th
e
cl
ass
i
f
ier
is
tr
ain
ed
in
o
r
d
er
t
o
p
r
e
d
ict
th
e
ex
ac
t
b
an
k
r
u
p
tc
y
.
T
h
e
p
r
ed
ictio
n
r
e
s
u
lt
s
f
o
r
b
an
k
r
u
p
tc
y
r
es
u
lt
s
ar
e
en
h
an
ce
d
b
y
e
m
p
lo
y
i
n
g
m
u
l
ti
m
o
d
al
G
A
b
ased
n
eu
r
al
n
et
w
o
r
k
.
C
o
r
r
elatio
n
m
atr
i
x
w
ill
ca
lcu
la
te
t
h
e
m
a
x
i
m
u
m
v
al
u
es
o
f
attr
ib
u
tes
o
n
t
h
e
b
asis
o
f
m
ap
p
in
g
tech
n
iq
u
e.
A
f
ter
t
h
at,
w
e
n
ee
d
to
g
iv
e
th
i
s
d
ata
to
th
e
class
if
i
er
,
s
h
o
w
n
in
t
h
e
F
ig
u
r
e
4
C
o
r
r
elatio
n
m
atr
ice
s
o
f
b
an
k
r
u
p
tc
y
d
ata
an
d
n
o
n
b
an
k
r
u
p
tc
y
d
ata
w
it
h
th
e
s
tatu
s
o
f
I
D
.
Fig
u
r
e
4
.
C
o
r
r
elatio
n
m
atr
ices
o
f
b
an
k
r
u
p
tc
y
d
ata
an
d
n
o
n
b
an
k
r
u
p
tc
y
d
ata
w
it
h
t
h
e
s
tat
u
s
o
f
I
D
(
0
an
d
1
)
Step
7
:
A
t
t
h
e
e
n
d
,
th
e
co
n
f
u
s
i
o
n
m
atr
ices
ar
e
ca
lc
u
lated
b
as
ed
o
n
T
P
FP
T
N
FN,
to
an
al
y
ze
th
e
p
er
f
o
r
m
a
n
ce
o
f
GA
-
ANN
class
i
f
ier
.
T
h
e
m
atr
ix
w
ill
s
h
o
w
t
h
e
b
an
k
r
u
p
tc
y
a
n
d
n
o
n
b
an
k
r
u
p
tc
y
d
ata
p
r
ed
ictio
n
ca
p
ac
it
y
o
f
th
e
clas
s
i
f
ier
alo
n
g
w
i
th
t
h
e
m
is
clas
s
if
icatio
n
r
ate.
T
P
m
ea
n
s
b
an
k
r
u
p
tc
y
d
ata
was
clas
s
i
f
ied
as
b
a
n
k
r
u
p
tc
y
a
n
d
n
o
n
b
an
k
r
u
p
tc
y
d
ata
w
a
s
cl
ass
i
f
ied
a
s
b
an
k
r
u
p
tc
y
.
FP
m
ea
n
s
b
an
k
r
u
p
tcy
d
ata
w
as
clas
s
i
f
ied
as
n
o
n
b
an
k
r
u
p
tc
y
,
FN
m
ea
n
s
n
o
n
b
an
k
r
u
p
tc
y
d
ata
w
a
s
class
i
f
ied
as
b
an
k
r
u
p
tc
y
.
T
N
m
ea
n
s
n
o
n
b
an
k
r
u
p
tc
y
d
ata
w
a
s
cla
s
s
i
f
ied
as
b
an
k
r
u
p
tc
y
an
d
b
an
k
r
u
p
tc
y
d
ata
w
a
s
class
if
ied
as
n
o
n
b
an
k
r
u
p
t
c
y
.
On
ce
t
h
e
p
r
o
p
o
s
ed
s
ch
e
m
e
is
d
esig
n
ed
,
th
e
p
er
f
o
r
m
a
n
ce
o
f
th
e
m
eth
o
d
w
ill
Evaluation Warning : The document was created with Spire.PDF for Python.
IJ
-
AI
I
SS
N:
2252
-
8938
P
r
ed
ictio
n
o
f b
a
n
kru
p
tcy
u
s
in
g
b
ig
d
a
ta
a
n
a
lytic
b
a
s
ed
o
n
f
u
z
z
y
c
-
mea
n
s
a
lg
o
r
ith
m
.
.
.
(
A
r
u
p
Gu
h
a
)
173
b
e
ev
alu
ated
b
ased
o
n
ac
cu
r
ac
y
,
p
r
ec
is
io
n
,
s
p
ec
if
icit
y
a
n
d
s
e
n
s
it
iv
i
t
y
,
s
h
o
w
n
in
t
h
e
Fi
g
u
r
e
5
.
Step
8
:
T
h
e
f
in
al
co
m
p
ar
i
s
o
n
h
as
b
ee
n
d
o
n
e
w
it
h
t
h
e
ex
is
t
in
g
m
e
th
o
d
s
to
k
n
o
w
th
e
ef
f
ec
t
iv
e
n
es
s
o
f
th
e
m
et
h
o
d
.
Fig
u
r
e
5
.
P
er
f
o
r
m
a
n
ce
ev
al
u
at
io
n
m
atr
i
x
4.
E
XP
E
R
I
M
E
NT
S AN
D
RE
S
UL
T
S
4
.
1
.
Resea
rc
h da
t
a
a
nd
ex
pe
ri
m
ent
s
T
h
e
d
ataset
co
m
p
r
is
e
o
f
f
in
an
cial
r
atio
s
o
f
s
e
v
er
al
s
m
a
ll
an
d
m
ed
i
u
m
s
ca
le
co
m
p
a
n
ies
f
r
o
m
1994
-
2
0
1
4
.
T
h
e
b
an
k
r
u
p
tc
y
an
d
n
o
n
-
b
an
k
r
u
p
tc
y
d
ata
s
tat
u
s
ar
e
s
h
o
w
n
i
n
t
h
e
F
ig
u
r
e
6
.
T
h
e
n
u
m
b
er
o
f
n
o
n
-
au
d
ited
co
m
p
a
n
ies
ar
e
f
o
u
n
d
co
m
p
ar
ati
v
el
y
h
i
g
h
er
t
h
a
t
th
e
to
tal
f
ir
m
s
.
T
h
e
d
ataset
w
a
s
s
p
lit
i
n
to
t
w
o
s
u
b
s
et
s
b
y
co
n
s
id
er
in
g
8
0
%
o
f
th
e
d
ata
f
o
r
tr
ain
in
g
d
ataset
w
h
ic
h
is
u
s
ed
to
d
ev
elo
p
u
n
d
er
s
a
m
p
li
n
g
m
et
h
o
d
f
o
r
d
ata
cla
s
s
b
ala
n
ci
n
g
a
n
d
2
0
%
f
o
r
a
v
al
id
atio
n
d
at
aset,
w
h
ic
h
i
s
a
r
r
an
g
ed
w
.
r
.
t
th
e
tr
ai
n
i
n
g
d
ata
d
is
tr
ib
u
tio
n
.
T
w
o
s
ta
g
e
s
el
ec
tio
n
p
r
o
ce
s
s
o
f
t
h
e
i
n
p
u
t
v
ar
iab
l
e
h
as
b
ee
n
ap
p
li
ed
b
ased
o
n
t
h
e
p
r
ev
io
u
s
m
et
h
o
d
[
1
,
3
]
.
T
h
e
ch
o
s
en
f
i
n
al
v
ar
iab
les
w
er
e
b
ase
d
o
n
th
e
va
r
ia
n
t te
s
t
an
d
th
e
s
e
v
ar
iab
le
w
er
e
u
s
ed
f
o
r
th
e
cr
ed
it
e
v
al
u
atio
n
o
f
t
h
e
s
elec
ted
co
m
p
an
ies.
T
h
e
m
o
d
el
is
i
m
p
le
m
e
n
ted
u
s
in
g
to
o
ls
p
y
t
h
o
n
3
.
6
an
d
An
ac
o
n
d
a
n
a
v
i
g
ato
r
.
Fig
u
r
e
6
.
Statu
s
o
f
b
an
k
r
u
p
tc
y
an
d
n
o
n
b
an
k
r
u
p
tc
y
4
.
2
.
Resul
t
a
nd
a
na
ly
s
is
E
f
f
ec
tiv
e
n
es
s
o
f
t
h
e
clu
s
ter
-
b
ased
GA
-
A
N
N
u
n
d
er
s
a
m
p
li
n
g
m
et
h
o
d
u
s
i
n
g
Fu
zz
y
C
m
ea
n
s
alg
o
r
ith
m
ap
p
lied
to
th
e
class
i
f
ier
w
a
s
b
ein
g
i
n
v
esti
g
ated
f
o
r
th
e
b
an
k
r
u
p
tc
y
p
r
ed
ictio
n
ap
p
licatio
n
.
Her
e
,
w
e
h
av
e
s
e
t
GA
to
s
ea
r
ch
t
h
e
cu
t
-
o
f
f
f
o
r
ea
ch
clu
s
ter
th
at
r
ep
r
esen
t
s
th
e
m
i
n
i
m
u
m
d
is
tan
ce
o
f
th
e
clu
s
ter
s
f
r
o
m
th
e
ce
n
tr
o
id
.
T
h
e
o
p
ti
m
izatio
n
tech
n
iq
u
e
s
ar
e
ap
p
lied
u
s
in
g
G
A
-
ANN,
th
at
h
as
led
to
a
cc
u
r
ate
p
r
ed
ictio
n
in
th
is
f
ea
t
u
r
e
m
atr
ix
.
I
n
t
h
e
class
i
f
icatio
n
m
o
d
el,
th
e
ap
p
lied
class
if
icat
io
n
alg
o
r
it
h
m
s
u
s
ed
w
er
e
Gen
etic
A
l
g
o
r
ith
m
b
ased
A
r
ti
f
icia
l
Neu
r
al
Net
w
o
r
k
s
,
lo
g
i
s
tic
R
e
g
r
ess
io
n
,
S
u
p
p
o
r
t
Vec
to
r
Ma
ch
in
es
an
d
Dec
is
io
n
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2252
-
8
9
3
8
IJ
-
AI
Vo
l.
8
,
No
.
2
,
J
u
n
e
20
1
9
:
1
68
–
1
7
4
174
T
r
ee
s
to
p
r
e
d
ict
b
an
k
r
u
p
tc
y
.
T
ested
Gen
etic
A
l
g
o
r
ith
m
b
ased
A
r
ti
f
icial
Neu
r
al
Net
w
o
r
k
s
w
er
e
f
o
u
n
d
ac
cu
r
ac
y
r
ate
o
f
7
8
.
2
1
%
w
ith
co
m
p
ar
is
o
n
to
ex
is
t
in
g
m
eth
o
d
ac
cu
r
ac
y
r
ate
an
d
s
h
o
w
ed
m
is
clas
s
if
icatio
n
r
ate
0
.
2
1
7
8
.
E
f
f
ec
ti
v
en
e
s
s
o
f
th
i
s
m
et
h
o
d
w
as
p
r
o
v
ed
b
y
co
m
p
a
r
in
g
its
ac
cu
r
ac
y
r
ate
w
it
h
t
h
e
r
esu
l
ts
o
f
e
x
is
tin
g
m
et
h
o
d
.
T
h
u
s
,
th
is
m
et
h
o
d
h
as
p
r
o
v
ed
ef
f
ec
ti
v
e
i
n
th
e
h
a
n
d
lin
g
o
f
s
u
c
h
i
m
b
a
lan
ce
d
at
aset
p
r
io
r
to
m
o
d
el
d
ev
elo
p
m
en
t
,
s
h
o
w
n
in
t
h
e
F
ig
u
r
e
7.
Fig
u
r
e
7
.
C
o
m
p
ar
is
o
n
o
f
m
o
d
el
f
o
r
ac
cu
r
ac
y
r
ate
5.
CO
NCLU
SI
O
N
T
h
is
s
tu
d
y
v
er
i
f
ied
th
e
e
f
f
ec
tiv
e
n
ess
o
f
th
e
p
r
o
p
o
s
ed
ap
p
r
o
ac
h
o
f
clu
s
ter
-
b
ased
u
n
d
er
-
s
a
m
p
li
n
g
u
s
i
n
g
Fu
zz
y
C
m
ea
n
s
alg
o
r
ith
m
in
o
r
d
er
to
o
p
tim
ize
G
A
-
ANN
f
o
r
ef
f
ec
t
iv
e
p
r
ed
ictio
n
o
f
b
an
k
r
u
p
tc
y
.
I
n
t
h
i
s
th
e
d
ata
is
s
tr
u
ctu
r
ed
b
y
c
lass
i
f
y
in
g
th
e
m
u
s
i
n
g
clu
s
t
er
in
g
tech
n
iq
u
e
a
n
d
p
er
f
o
r
m
i
n
g
s
i
m
u
lta
n
eo
u
s
o
p
tim
izatio
n
f
o
r
th
e
A
NN
m
o
d
el.
T
h
is
m
et
h
o
d
h
as
led
to
th
e
ef
f
ec
ti
v
e
n
es
s
o
f
th
e
cla
s
s
i
f
ie
r
an
d
d
ec
r
ea
s
in
g
th
e
d
ata
i
m
b
alan
ce
r
ate
at
t
h
e
s
a
m
e
ti
m
e.
T
h
e
ex
p
er
i
m
en
tal
r
e
s
u
lt
s
h
o
w
ed
a
n
ac
c
u
r
ac
y
o
f
7
8
.
2
%
as
co
m
p
ar
ed
to
th
e
ex
i
s
ti
n
g
m
et
h
o
d
s
.
RE
F
E
R
E
NC
E
S
[1
]
T
a
m
b
e
,
P
.
(
2
0
1
4
).
Big
d
a
ta i
n
v
e
st
m
e
n
t,
sk
il
ls,
a
n
d
f
ir
m
v
a
lu
e
.
M
a
n
a
g
e
m
e
n
t
S
c
ien
c
e
,
6
0
(6
),
1
4
5
2
-
1
4
6
9
.
[2
]
Kim
,
K.
J.
,
&
A
h
n
,
H.
(2
0
1
2
).
A
c
o
rp
o
ra
te
c
re
d
it
ra
ti
n
g
m
o
d
e
l
u
s
in
g
m
u
lt
i
-
c
las
s
su
p
p
o
rt
v
e
c
to
r
m
a
c
h
in
e
s
w
it
h
a
n
o
rd
i
n
a
l
p
a
irw
ise
p
a
rti
ti
o
n
in
g
a
p
p
r
o
a
c
h
.
Co
m
p
u
ters
&
Op
e
ra
ti
o
n
s R
e
se
a
rc
h
,
3
9
(8
)
,
1
8
0
0
-
1
8
1
1
[3
]
L
e
,
T
.
,
Le
S
o
n
,
H
.
,
V
o
,
M
.
,
L
e
e
,
M
.
,
&
Ba
ik
,
S
.
(2
0
1
8
).
A
c
lu
ste
r
-
b
a
se
d
b
o
o
st
in
g
a
lg
o
rit
h
m
f
o
r
b
a
n
k
ru
p
tcy
p
re
d
ictio
n
i
n
a
h
ig
h
ly
i
m
b
a
lan
c
e
d
d
a
tas
e
t.
S
y
m
m
e
tr
y
,
1
0
(7
),
2
5
0
.
[4
]
Kim
,
H.
J.,
Jo
,
N.
O.,
&
S
h
in
,
K.
S
.
(2
0
1
6
).
Op
ti
m
iza
ti
o
n
o
f
c
l
u
ste
r
-
b
a
se
d
e
v
o
lu
ti
o
n
a
ry
u
n
d
e
rsa
m
p
li
n
g
f
o
r
th
e
a
rti
f
icia
l
n
e
u
ra
l
n
e
tw
o
rk
s in
c
o
rp
o
ra
te b
a
n
k
ru
p
tcy
p
re
d
ictio
n
.
Ex
p
e
r
t
S
y
ste
m
s
w
it
h
A
p
p
li
c
a
ti
o
n
s,
5
9
,
2
2
6
-
2
3
4
.
[5
]
S
o
n
g
,
A
.
,
&
X
u
,
Q.
(
2
0
1
8
)
.
Im
b
a
lan
c
e
d
Da
ta
Clas
sif
ic
a
ti
o
n
Ba
se
d
o
n
M
BCDK
-
m
e
a
n
s
Un
d
e
rsa
m
p
li
n
g
a
n
d
G
A
-
A
N
N.
In
In
ter
n
a
ti
o
n
a
l
Co
n
f
e
re
n
c
e
o
n
A
rti
f
icia
l
Ne
u
ra
l
Ne
t
w
o
rk
s (
p
p
.
3
4
9
-
3
5
8
).
S
p
rin
g
e
r,
C
h
a
m
.
[6
]
Ye
n
,
S
.
J.
,
&
L
e
e
,
Y.
S
.
(2
0
0
9
)
.
C
lu
ste
r
-
b
a
se
d
u
n
d
e
r
-
sa
m
p
li
n
g
a
p
p
r
o
a
c
h
e
s
f
o
r
i
m
b
a
lan
c
e
d
d
a
ta
d
istri
b
u
ti
o
n
s.
Ex
p
e
rt
S
y
st
e
m
s
w
it
h
A
p
p
li
c
a
ti
o
n
s,
3
6
(3
)
,
5
7
1
8
-
5
7
2
7
[7
]
Ka
n
g
,
P
.
,
Ch
o
,
S
.
,
&
M
a
c
L
a
c
h
lan
,
D.
L
.
(2
0
1
2
).
Im
p
ro
v
e
d
re
sp
o
n
se
m
o
d
e
li
n
g
b
a
se
d
o
n
c
lu
ste
rin
g
,
u
n
d
e
r
-
sa
m
p
li
n
g
,
a
n
d
e
n
se
m
b
le.
Ex
p
e
rt
S
y
s
te
m
w
it
h
A
p
p
li
c
a
ti
o
n
s,
3
9
(
8
),
6
7
3
8
-
6
7
5
3
.
[8
]
Kh
o
sh
g
o
f
taa
r,
T
.
M
.
,
S
e
li
y
a
,
N.
,
&
Dro
w
n
,
D.
J.
(2
0
1
0
)
.
Ev
o
l
u
ti
o
n
a
ry
d
a
ta
a
n
a
l
y
sis
f
o
r
th
e
c
las
s
i
m
b
a
lan
c
e
p
ro
b
lem
.
In
telli
g
e
n
t
Da
ta A
n
a
l
y
si
s,
1
4
(1
),
6
9
-
88
[9
]
G
a
r
c
ía,
S
.
,
&
H
e
rre
ra
,
F
.
(2
0
0
9
).
Ev
o
lu
ti
o
n
a
ry
u
n
d
e
rsa
m
p
li
n
g
f
o
r
c
las
sif
ic
a
ti
o
n
w
it
h
i
m
b
a
lan
c
e
d
d
a
tas
e
ts:
P
ro
p
o
sa
ls
a
n
d
tax
o
n
o
m
y
.
Ev
o
lu
ti
o
n
a
ry
Co
m
p
u
tati
o
n
,
1
7
(3
)
,
2
7
5
-
3
0
6
.
[1
0
]
Ch
o
w
,
J.
C.
(
2
0
1
8
).
A
n
a
l
y
sis o
f
F
in
a
n
c
ial
Cre
d
it
Risk
Us
in
g
M
a
c
h
i
n
e
L
e
a
rn
in
g
.
a
rX
iv
p
re
p
r
in
t
a
rX
iv
:1
8
0
2
.
0
5
3
2
6
.
[1
1
]
V
a
n
n
u
c
c
i,
M
.
,
&
Co
ll
a
,
V
.
(
2
0
1
7
).
G
e
n
e
ti
c
A
lg
o
rit
h
m
s
Ba
s
e
d
Re
sa
m
p
li
n
g
f
o
r
th
e
Clas
sif
ica
ti
o
n
o
f
Un
b
a
lan
c
e
d
Da
tas
e
ts.
In
In
ter
n
a
ti
o
n
a
l
Co
n
fer
e
n
c
e
o
n
I
n
telli
g
e
n
t
De
c
isio
n
T
e
c
h
n
o
l
o
g
ies
(
p
p
.
2
3
-
3
2
).
S
p
ri
n
g
e
r,
Ch
a
m
.
[1
2
]
Do
n
g
,
S
.
,
&
W
u
,
Y.
(2
0
1
8
,
Ju
l
y
).
A
g
e
n
e
ti
c
a
lg
o
rit
h
m
-
b
a
s
e
d
a
p
p
r
o
a
c
h
f
o
r
c
las
s
-
i
m
b
a
lan
c
e
d
le
a
rn
in
g
.
In
T
h
ir
d
In
ter
n
a
t
io
n
a
l
W
o
rk
sh
o
p
o
n
P
a
tt
e
rn
Rec
o
g
n
it
io
n
(Vo
l.
1
0
8
2
8
,
p
.
1
0
8
2
8
1
D).
In
ter
n
a
ti
o
n
a
l
S
o
c
iety
f
o
r
Op
ti
c
s
a
n
d
P
h
o
t
o
n
ics
.
[1
3
]
Qin
,
J.,
F
u
,
W
.
,
G
a
o
,
H.,
&
Zh
e
n
g
,
W
.
X
.
(2
0
1
7
).
Distr
ib
u
ted
$
k
$
-
m
e
a
n
s
a
lg
o
rit
h
m
a
n
d
f
u
z
z
y
$
c
$
-
m
e
a
n
s
a
lg
o
rit
h
m
f
o
r
se
n
so
r
n
e
tw
o
rk
s
b
a
se
d
o
n
m
u
lt
i
a
g
e
n
t
c
o
n
se
n
su
s
th
e
o
ry
.
IEE
E
tra
n
sa
c
ti
o
n
s
o
n
c
y
b
e
rn
e
ti
c
s
,
47
(3
)
,
772
-
7
8
3
.
Evaluation Warning : The document was created with Spire.PDF for Python.