I
nte
rna
t
io
na
l J
o
urna
l o
f
I
nfo
rm
a
t
ics a
nd
Co
m
m
un
ica
t
io
n T
ec
hn
o
lo
g
y
(
I
J
-
I
CT
)
Vo
l.
1
5
,
No
.
4
,
D
e
c
em
b
er
20
2
6
,
p
p
.
1
6
1
4
~
1
6
2
3
I
SS
N:
2252
-
8
7
7
6
,
DOI
:
1
0
.
1
1
5
9
1
/iji
ct
.
v
1
5
i
4
.
pp
1
6
1
4
-
1
6
2
3
1614
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ij
ict.
ia
esco
r
e.
co
m
Exploring
ef
fec
tive fe
a
tu
re sele
ction
methods
in
log
istic
regres
sio
n f
o
r cr
e
dit
sco
ring
mo
del
s in XYZ
bank
B
rig
it
t
a
P
er
m
a
dh
i
,
Dev
i F
it
r
ia
na
h
C
o
m
p
u
t
e
r
S
c
i
e
n
c
e
D
e
p
a
r
t
m
e
n
t
,
B
I
N
U
S
G
r
a
d
u
a
t
e
P
r
o
g
r
a
m
–
M
a
s
t
e
r
o
f
C
o
m
p
u
t
e
r
S
c
i
e
n
c
e
,
B
i
n
a
N
u
sa
n
t
a
r
a
U
n
i
v
e
r
si
t
y
,
J
a
k
a
r
t
a
,
I
n
d
o
n
e
s
i
a
Art
icle
I
nfo
AB
S
T
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
No
v
1
6
,
2
0
2
5
R
ev
is
ed
J
u
n
2
5
,
2
0
2
6
Acc
ep
ted
Au
g
2
8
,
2
0
2
6
Th
is
re
se
a
rc
h
e
x
p
l
o
re
s
o
p
ti
m
ize
d
v
a
riab
le
se
lec
ti
o
n
in
c
re
d
it
sc
o
ri
n
g
m
o
d
e
ls,
c
o
m
p
a
rin
g
wra
p
p
e
r,
e
m
b
e
d
d
e
d
,
n
o
n
e
,
a
n
d
h
y
b
r
id
a
p
p
r
o
a
c
h
e
s
.
Lo
g
isti
c
re
g
re
ss
io
n
is
a
p
p
li
e
d
to
m
o
d
e
l
t
h
e
d
e
p
e
n
d
e
n
c
e
o
f
e
x
p
la
n
a
to
ry
v
a
riab
les
o
n
th
e
targ
e
t
v
a
riab
le
wh
il
e
m
it
ig
a
ti
n
g
u
n
d
e
rfit
ti
n
g
,
o
v
e
rfi
tt
in
g
,
a
n
d
c
o
m
p
u
tati
o
n
a
l
c
h
a
ll
e
n
g
e
s.
F
e
a
tu
re
se
lec
ti
o
n
m
e
th
o
d
s
h
e
l
p
m
it
i
g
a
te
th
e
se
issu
e
s
b
y
re
d
u
c
in
g
d
ime
n
si
o
n
a
li
t
y
a
n
d
imp
r
o
v
i
n
g
m
o
d
e
l
g
e
n
e
ra
li
z
a
ti
o
n
.
T
h
e
Hy
b
ri
d
m
o
d
e
l
a
c
h
iev
e
d
th
e
b
e
st
o
v
e
ra
ll
p
e
rfo
rm
a
n
c
e
with
a
c
c
u
ra
c
y
=
0
.
9
8
4
0
o
n
t
h
e
tes
t
se
t
a
n
d
a
c
c
u
ra
c
y
=
0
.
9
8
3
9
o
n
th
e
train
i
n
g
se
t
(
tes
t
-
train
d
iffere
n
c
e
=
-
0
.
0
0
0
1
),
a
s
we
ll
a
s
th
e
b
e
st
r
e
c
a
ll
=
0
.
9
9
9
3
,
F
1
-
sc
o
re
=
0
.
9
8
4
2
,
a
n
d
th
e
lea
st n
u
m
b
e
r
o
f
fa
lse
p
o
siti
v
e
s (F
P
=
4
7
)
a
n
d
fa
lse
n
e
g
a
ti
v
e
s (F
N = 1
),
wh
ich
su
g
g
e
sts
g
o
o
d
p
re
d
icti
v
e
a
b
il
it
y
a
n
d
sta
b
il
it
y
wit
h
th
e
fe
we
st
m
isc
las
sifica
ti
o
n
s.
Th
e
wra
p
p
e
r
a
n
d
e
m
b
e
d
d
e
d
m
o
d
e
ls
h
a
v
e
l
o
we
r
a
c
c
u
ra
c
ies
o
f
0
.
9
7
9
3
a
n
d
0
.
9
8
3
0
,
re
sp
e
c
ti
v
e
ly
,
a
n
d
m
il
d
o
v
e
rfi
tt
in
g
.
Th
e
b
a
se
li
n
e
m
o
d
e
l
with
n
o
fe
a
tu
re
se
lec
ti
o
n
p
e
rfo
rm
e
d
th
e
wo
rst
i
n
term
s
o
f
b
o
t
h
a
c
c
u
ra
c
y
a
n
d
F
1
-
sc
o
re
(a
c
c
u
ra
c
y
=
0
.
5
0
0
0
,
F
1
-
sc
o
re
=
0
.
6
6
6
7
).
I
n
c
o
n
c
lu
si
o
n
,
t
h
e
re
su
lt
s
i
n
d
ica
te
th
a
t
t
h
e
c
o
m
b
in
a
ti
o
n
o
f
wra
p
p
e
r
a
n
d
e
m
b
e
d
d
e
d
a
p
p
ro
a
c
h
e
s
is
th
e
m
o
st
a
c
c
u
ra
te,
sta
b
le,
a
n
d
g
e
n
e
ra
li
sa
b
le
m
o
d
e
l
fo
r
c
re
d
it
sc
o
ri
n
g
tas
k
s.
K
ey
w
o
r
d
s
:
Acc
u
r
ac
y
C
r
ed
it sco
r
in
g
Featu
r
e
s
elec
tio
n
Hy
b
r
id
m
o
d
el
L
o
g
is
tic
r
eg
r
ess
io
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
:
B
r
ig
itta Per
m
ad
h
i
C
o
m
p
u
ter
Scien
ce
Dep
ar
tm
e
n
t,
B
I
NUS
Gr
ad
u
ate
Pro
g
r
a
m
–
Ma
s
ter
o
f
C
o
m
p
u
ter
Scien
ce
B
in
a
Nu
s
an
tar
a
Un
iv
er
s
ity
J
ak
ar
ta,
I
n
d
o
n
esia
E
m
ail: b
r
ig
itta.p
er
m
a
d
h
i@
b
in
u
s
.
ac
.
id
1.
I
NT
RO
D
UCT
I
O
N
I
n
to
d
ay
'
s
d
ata
-
d
r
iv
e
n
wo
r
l
d
,
elig
ib
ilit
y
s
co
r
in
g
m
o
d
els
p
la
y
a
c
r
u
cial
r
o
le
in
s
u
p
p
o
r
tin
g
in
d
iv
id
u
als
an
d
o
r
g
an
izatio
n
s
in
m
ak
i
n
g
well
-
in
f
o
r
m
ed
d
ec
is
io
n
s
[
1
]
.
E
lig
ib
ilit
y
s
co
r
in
g
is
a
f
o
u
n
d
atio
n
al
p
r
o
ce
s
s
f
o
r
ev
alu
atin
g
o
b
jects
b
ased
o
n
p
er
tin
en
t
v
ar
iab
le
s
,
allo
win
g
f
o
r
th
e
p
r
e
d
ictio
n
o
f
s
p
ec
if
ic
ch
ar
ac
ter
is
tics
[
2
]
.
T
h
is
ca
p
ab
ilit
y
en
ab
les
b
u
s
in
ess
es
an
d
in
d
iv
id
u
als
to
as
s
ess
r
is
k
s
an
d
m
ak
e
well
in
f
o
r
m
ed
ch
o
ices
[
3
]
.
I
t h
as a
p
p
licatio
n
s
ac
r
o
s
s
v
ar
i
o
u
s
s
ec
to
r
s
,
in
clu
d
in
g
f
in
an
ce
,
h
ea
lth
ca
r
e,
m
a
r
k
etin
g
,
an
d
i
n
s
u
r
an
ce
[
4
]
.
T
h
is
m
eth
o
d
ass
ig
n
s
n
u
m
er
ical
s
co
r
es
to
o
b
jects
f
o
r
c
o
m
p
ar
is
o
n
,
class
if
icatio
n
,
an
d
d
ec
is
io
n
-
m
ak
in
g
,
m
ak
in
g
it
ess
en
tial
in
f
ield
s
r
eq
u
ir
in
g
o
b
jectiv
e
an
aly
s
is
.
I
n
cr
ed
it
s
co
r
in
g
,
p
r
e
d
ictiv
e
tech
n
iq
u
es
s
u
ch
as
lo
g
is
tic
r
e
g
r
ess
io
n
,
r
an
d
o
m
f
o
r
est
an
d
d
ec
is
io
n
tr
ee
s
an
aly
ze
d
ata
t
o
ass
es
s
cr
ed
it
r
is
k
,
p
r
o
d
u
cin
g
s
co
r
es th
at
i
n
d
icate
lev
els o
f
r
is
k
,
q
u
ality
o
r
p
r
o
b
a
b
ilit
y
[
2
]
.
A
k
ey
ch
allen
g
e
in
cr
ed
it
s
co
r
in
g
is
o
v
er
f
itti
n
g
,
wh
ich
o
cc
u
r
s
wh
en
a
m
o
d
el
b
ec
o
m
es
to
o
t
ailo
r
ed
to
th
e
tr
ain
in
g
d
ata,
r
e
d
u
cin
g
it
s
ab
ilit
y
to
g
en
e
r
alize
to
n
e
w
d
ata,
lead
in
g
to
p
o
o
r
r
ea
l
-
wo
r
ld
p
er
f
o
r
m
a
n
ce
.
C
o
n
v
er
s
ely
,
u
n
d
e
r
f
itti
n
g
o
cc
u
r
s
wh
en
th
e
m
o
d
el
is
to
o
s
im
p
lis
tic
to
r
ec
o
g
n
ize
m
ea
n
i
n
g
f
u
l
d
ata
p
atter
n
s
,
r
esu
ltin
g
in
wea
k
p
r
ed
ictiv
e
p
er
f
o
r
m
a
n
ce
o
n
b
o
th
t
r
ain
in
g
a
n
d
te
s
tin
g
s
ets [
5
]
.
T
o
a
d
d
r
ess
th
ese
is
s
u
es,
f
ea
tu
r
e
s
elec
tio
n
m
eth
o
d
s
ar
e
ess
en
tial f
o
r
ef
f
ec
tiv
ely
r
ed
u
cin
g
f
ea
tu
r
e
d
im
en
s
io
n
ality
[
6
]
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J I
n
f
&
C
o
m
m
u
n
T
ec
h
n
o
l
I
SS
N:
2252
-
8
7
7
6
E
xp
lo
r
in
g
effec
tive
fea
tu
r
e
s
elec
tio
n
meth
o
d
s
in
lo
g
is
t
ic
r
eg
r
ess
io
n
fo
r
cred
it
…
(
B
r
ig
itta
P
erma
d
h
i
)
1615
Featu
r
e
s
elec
tio
n
m
o
d
els ar
e
v
alu
ab
le
f
o
r
d
ata
an
al
y
s
ts
in
s
el
ec
tin
g
th
e
m
o
s
t sig
n
if
ican
t v
ar
iab
les [
7
]
.
B
y
s
y
s
tem
atica
lly
s
e
lectin
g
o
n
ly
th
e
m
o
s
t
r
elev
a
n
t
f
ea
tu
r
e
s
,
f
ea
tu
r
e
s
elec
tio
n
en
h
an
ce
s
m
o
d
el
ef
f
icien
c
y
,
s
im
p
lifie
s
co
m
p
u
tatio
n
s
,
an
d
r
ed
u
ce
s
o
v
er
f
itti
n
g
r
is
k
,
lea
d
in
g
to
m
o
r
e
i
n
ter
p
r
eta
b
le
a
n
d
co
m
p
u
tatio
n
ally
ef
f
icien
t c
r
ed
it sco
r
in
g
m
o
d
els [
8
]
.
T
h
e
wr
ap
p
er
m
eth
o
d
f
o
llo
w
s
an
iter
ativ
e
p
r
o
ce
s
s
wh
ic
h
is
i
n
itializin
g
a
f
ea
tu
r
e
s
u
b
s
et,
tr
ain
in
g
th
e
m
o
d
el,
ev
al
u
atin
g
its
p
er
f
o
r
m
an
ce
,
u
p
d
atin
g
th
e
s
u
b
s
et,
an
d
r
ep
ea
tin
g
t
h
e
cy
cle
u
n
til
a
s
to
p
p
in
g
co
n
d
itio
n
is
m
et
[
9
]
.
T
h
e
em
b
e
d
d
ed
m
eth
o
d
in
te
g
r
ates
f
ea
tu
r
e
s
elec
tio
n
with
in
th
e
m
o
d
el
tr
ain
in
g
p
r
o
ce
s
s
,
allo
win
g
th
e
alg
o
r
ith
m
to
lear
n
an
d
ass
ig
n
h
ig
h
er
im
p
o
r
tan
ce
to
th
e
m
o
s
t
r
elev
an
t
f
ea
tu
r
es
wh
ile
m
in
im
izin
g
th
e
in
f
lu
en
ce
o
f
less
s
ig
n
if
ican
t
o
n
es
[
8
]
.
Hen
ce
,
we
p
r
o
p
o
s
ed
a
n
ew
h
y
b
r
id
ap
p
r
o
ac
h
,
wh
ich
co
m
b
in
es
m
u
ltip
le
s
elec
tio
n
tech
n
iq
u
es
to
le
v
er
ag
e
t
h
eir
r
esp
ec
tiv
e
s
tr
en
g
th
s
,
e
n
h
an
cin
g
m
o
d
el
in
ter
p
r
etab
ilit
y
,
p
er
f
o
r
m
an
ce
,
a
n
d
m
itig
atin
g
o
v
er
f
itti
n
g
[
1
0
]
.
W
h
ile
h
y
b
r
id
f
ea
tu
r
e
s
elec
tio
n
m
eth
o
d
s
s
u
ch
as
L
ass
o
c
o
m
b
in
ed
with
s
eq
u
e
n
tial
f
ea
tu
r
e
s
elec
tio
n
h
av
e
b
ee
n
p
r
ev
io
u
s
ly
ex
p
lo
r
ed
,
th
i
s
s
tu
d
y
d
o
es
n
o
t
aim
to
p
r
o
p
o
s
e
a
f
u
n
d
am
en
tall
y
n
ew
alg
o
r
ith
m
.
I
n
s
tead
,
th
e
co
n
tr
ib
u
tio
n
o
f
th
i
s
r
esear
ch
lies
in
th
e
co
n
tex
tu
al
ad
ap
tatio
n
an
d
em
p
ir
ical
v
a
lid
atio
n
o
f
a
h
y
b
r
id
f
ea
tu
r
e
s
elec
tio
n
s
tr
ateg
y
o
n
a
lar
g
e
-
s
ca
le,
r
ea
l
-
wo
r
ld
b
an
k
i
n
g
d
ataset
[
1
0
]
.
Un
lik
e
p
r
io
r
s
tu
d
ies
th
at
f
o
cu
s
o
n
m
eth
o
d
o
l
o
g
ical
n
o
v
elty
,
th
is
s
tu
d
y
e
m
p
h
asizes
h
o
w
th
e
in
te
g
r
atio
n
o
f
em
b
e
d
d
ed
an
d
wr
a
p
p
er
m
et
h
o
d
s
ca
n
b
e
ef
f
ec
tiv
ely
a
p
p
lied
to
h
ig
h
-
d
i
m
en
s
io
n
al
an
d
im
b
alan
ce
d
f
i
n
an
cial
d
ata.
T
h
e
r
esu
lts
d
e
m
o
n
s
tr
ate
th
at
s
u
ch
ad
ap
tatio
n
lead
s
t
o
a
s
tab
le,
in
ter
p
r
etab
le,
a
n
d
g
en
er
aliza
b
le
m
o
d
el,
w
h
ich
is
cr
itical
f
o
r
p
r
ac
tical
cr
ed
it
s
co
r
in
g
ap
p
licatio
n
s
.
L
o
g
is
tic
R
eg
r
ess
io
n
is
r
eta
in
ed
in
th
is
s
tu
d
y
d
u
e
to
its
s
t
r
o
n
g
in
ter
p
r
etab
ilit
y
an
d
alig
n
m
en
t
with
r
eg
u
lato
r
y
r
eq
u
ir
e
m
en
ts
co
m
m
o
n
ly
f
o
u
n
d
in
f
i
n
an
cial
in
s
titu
tio
n
s
[
1
1
]
.
I
n
cr
ed
it
s
co
r
in
g
ap
p
licatio
n
s
,
m
o
d
els
m
u
s
t
n
o
t
o
n
ly
ac
h
iev
e
h
ig
h
p
r
ed
ictiv
e
p
er
f
o
r
m
an
c
e
b
u
t
also
p
r
o
v
id
e
tr
an
s
p
ar
e
n
t
an
d
e
x
p
lain
ab
l
e
d
ec
is
io
n
-
m
ak
in
g
p
r
o
ce
s
s
es.
L
o
g
is
tic
R
eg
r
ess
io
n
en
ab
les
p
r
ac
titi
o
n
er
s
to
clea
r
l
y
u
n
d
er
s
ta
n
d
th
e
r
elatio
n
s
h
ip
b
etwe
en
in
p
u
t
v
ar
iab
les an
d
cr
ed
it r
is
k
,
m
ak
in
g
it su
itab
le
f
o
r
au
d
it a
n
d
co
m
p
lian
ce
p
u
r
p
o
s
es [
1
2
]
.
Fu
r
th
er
m
o
r
e
,
f
ea
tu
r
e
s
elec
tio
n
is
n
o
t
o
n
ly
im
p
o
r
tan
t
f
o
r
i
m
p
r
o
v
i
n
g
m
o
d
el
p
e
r
f
o
r
m
an
c
e
b
u
t
also
p
r
o
v
id
es
s
ig
n
if
ican
t
b
u
s
in
ess
v
alu
e
[
5
]
.
B
y
r
ed
u
cin
g
th
e
n
u
m
b
er
o
f
in
p
u
t
v
ar
iab
les,
it
s
im
p
lifie
s
m
o
d
el
in
ter
p
r
etatio
n
f
o
r
s
tak
eh
o
ld
er
s
an
d
s
u
p
p
o
r
ts
r
e
g
u
lato
r
y
tr
a
n
s
p
ar
en
cy
.
I
n
a
d
d
itio
n
,
f
ewe
r
v
a
r
iab
les
r
ed
u
ce
d
ata
co
llectio
n
,
s
to
r
ag
e
,
an
d
p
r
o
c
ess
in
g
co
s
ts
,
m
ak
in
g
th
e
m
o
d
el
m
o
r
e
ef
f
icie
n
t
an
d
p
r
ac
tical
f
o
r
r
ea
l
-
wo
r
l
d
im
p
lem
en
tatio
n
in
b
an
k
i
n
g
en
v
ir
o
n
m
e
n
ts
.
T
h
is
s
tu
d
y
aim
s
to
o
p
tim
ize
f
ea
tu
r
e
s
elec
tio
n
in
cr
ed
it
s
co
r
in
g
m
o
d
els
u
s
in
g
d
ata
f
r
o
m
B
an
k
XYZ
,
f
o
cu
s
in
g
o
n
im
p
r
o
v
in
g
m
o
d
el
ac
cu
r
ac
y
an
d
ef
f
i
cien
c
y
.
T
h
e
r
esear
ch
m
eth
o
d
in
cl
u
d
es
d
ata
co
llectio
n
,
s
ettin
g
o
b
jectiv
e,
clea
n
s
in
g
d
ata,
f
e
atu
r
e
s
elec
tio
n
,
d
ata
s
p
litt
in
g
,
in
p
u
ttin
g
p
r
e
p
r
o
ce
s
s
ed
d
a
ta
in
to
t
h
e
m
o
d
el,
ev
alu
atio
n
a
n
d
v
alid
atio
n
[
1
1
]
.
T
h
is
s
tu
d
y
will
co
n
tr
ib
u
te
t
o
ad
v
an
ci
n
g
f
ea
t
u
r
e
s
elec
tio
n
a
n
d
em
p
h
asize
th
eir
im
p
o
r
tan
ce
.
2.
RE
S
E
ARCH
M
E
T
H
O
D
T
h
e
r
esear
ch
p
r
o
ce
s
s
b
eg
i
n
s
with
d
ata
co
llectio
n
,
f
o
llo
wed
b
y
d
ef
in
in
g
th
e
m
o
d
el’
s
o
b
jectiv
e.
Nex
t,
d
ata
clea
n
in
g
is
p
er
f
o
r
m
ed
to
en
s
u
r
e
th
e
q
u
ality
an
d
co
n
s
is
ten
cy
o
f
th
e
d
ataset.
Featu
r
e
s
elec
tio
n
is
th
en
ap
p
lied
to
id
en
tif
y
th
e
m
o
s
t
s
i
g
n
if
ican
t
v
ar
ia
b
les.
I
n
th
is
s
tu
d
y
,
f
o
u
r
f
ea
t
u
r
e
s
elec
tio
n
s
tr
ateg
ies
ar
e
em
p
lo
y
ed
:
W
r
ap
p
er
,
E
m
b
e
d
d
ed
,
Hy
b
r
id
,
a
n
d
a
b
aselin
e
with
n
o
f
e
atu
r
e
s
elec
tio
n
,
ch
o
s
en
to
r
e
p
r
esen
t
a
r
an
g
e
o
f
co
m
p
u
tatio
n
al
c
o
m
p
lex
ity
a
n
d
g
en
er
aliza
tio
n
ab
ilit
y
.
T
o
m
in
im
ize
p
o
ten
tial
b
ias
an
d
en
h
an
ce
r
ep
r
o
d
u
cib
ilit
y
,
th
e
d
ataset
was
r
an
d
o
m
ly
d
iv
i
d
ed
in
to
tr
ain
in
g
an
d
test
in
g
s
u
b
s
ets.
T
h
e
lo
g
is
tic
r
eg
r
ess
io
n
m
o
d
el
was
f
itted
u
s
in
g
th
e
tr
ain
in
g
d
ata
an
d
s
u
b
s
eq
u
en
tly
e
v
alu
ated
o
n
th
e
test
s
et
to
ass
ess
its
p
r
ed
i
ctiv
e
p
er
f
o
r
m
a
n
ce
.
E
ac
h
f
ea
tu
r
e
s
elec
tio
n
ap
p
r
o
ac
h
is
ass
ess
ed
u
s
in
g
s
tan
d
ar
d
class
if
icatio
n
m
etr
ics,
i
n
clu
d
in
g
ac
cu
r
ac
y
,
p
r
ec
is
io
n
,
r
e
ca
ll,
an
d
F1
-
s
co
r
e
[
1
3
]
.
T
h
e
r
esu
lts
o
b
tain
ed
f
r
o
m
th
ese
ev
alu
atio
n
s
ar
e
an
aly
ze
d
an
d
co
m
p
ar
ed
to
d
eter
m
in
e
th
e
r
elativ
e
ef
f
ec
tiv
en
ess
o
f
ea
ch
f
ea
tu
r
e
s
elec
tio
n
m
eth
o
d
.
A
ce
n
tr
al
h
y
p
o
th
esis
o
f
th
is
s
tu
d
y
is
th
at
th
e
H
y
b
r
id
m
eth
o
d
will
y
ield
s
u
p
er
io
r
r
esu
lt
s
b
y
c
o
m
b
in
in
g
s
tatis
tical
r
elev
an
ce
an
d
m
o
d
el
-
s
p
ec
if
ic
r
ef
in
em
en
t.
T
h
e
f
in
al
s
tep
in
v
o
lv
es
v
alid
atin
g
th
e
m
o
d
el'
s
p
er
f
o
r
m
a
n
ce
o
n
n
ew
d
ata
[
1
4
]
.
Fig
u
r
e
1
illu
s
tr
ates
th
e
en
tire
p
r
o
ce
s
s
.
2
.
1
.
Da
t
a
c
o
llect
io
n
T
h
e
f
ir
s
t
s
tep
in
v
o
lv
es
o
b
tain
in
g
th
e
d
ataset
f
o
r
th
e
elig
ib
i
lity
s
co
r
in
g
m
o
d
el.
I
n
ad
d
itio
n
to
d
ata
ac
q
u
is
itio
n
,
th
is
p
h
ase
also
in
v
o
lv
es
id
en
tify
i
n
g
th
e
d
ep
e
n
d
en
t
v
ar
iab
le
(
)
a
n
d
in
d
ep
en
d
en
t
v
ar
iab
les
(
)
in
th
e
p
r
ed
ictiv
e
m
o
d
el.
I
t
is
e
s
s
en
tial
to
en
s
u
r
e
th
at
th
e
d
ataset
co
n
tain
s
r
elev
an
t
f
ea
tu
r
e
s
n
ec
ess
ar
y
f
o
r
p
r
ed
ictin
g
th
e
d
esire
d
o
u
tco
m
es
[
1
]
.
I
n
th
is
s
tu
d
y
,
d
ata
is
o
b
tain
ed
f
r
o
m
B
an
k
XYZ
an
d
p
er
tain
s
to
c
r
ed
it
ap
p
r
o
v
als,
m
ak
in
g
it
a
cr
ed
it
s
co
r
in
g
m
o
d
el
[
1
5
]
.
I
n
th
is
m
o
d
el,
th
e
tar
g
et
v
a
r
iab
le
(
)
is
cr
ed
it
ap
p
r
o
v
al
d
ec
is
io
n
an
d
th
e
o
th
er
co
lu
m
n
s
b
ec
o
m
e
th
e
in
d
ep
en
d
en
t v
a
r
iab
les (
).
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8
7
7
6
I
n
t J I
n
f
&
C
o
m
m
u
n
T
ec
h
n
o
l
,
Vo
l.
1
5
,
No
.
4
,
D
e
c
em
b
er
20
2
6
:
1
6
1
4
-
1
6
2
3
1616
Fig
u
r
e
1
.
R
esear
ch
m
eth
o
d
2
.
2
.
Set
t
ing
o
bje
ct
iv
e
I
n
th
is
s
tep
,
th
e
class
if
icatio
n
cr
iter
ia
f
o
r
cr
ed
itwo
r
t
h
in
ess
ar
e
d
ef
in
e
d
.
Sp
ec
if
ically
,
cr
ed
it
s
co
r
es
o
f
3
,
4
,
an
d
5
ar
e
ca
teg
o
r
ized
as
"Bad
,
"
wh
ile
s
co
r
es
o
f
1
an
d
2
ar
e
co
n
s
id
er
e
d
"Go
o
d
"
.
T
h
ese
Go
o
d
/B
ad
class
if
icatio
n
s
s
er
v
e
as
th
e
tar
g
et
v
ar
iab
le
f
o
r
tr
ain
in
g
th
e
m
o
d
el.
T
h
e
d
ef
in
itio
n
s
o
f
g
o
o
d
an
d
b
ad
is
f
lex
ib
le
an
d
ca
n
b
e
ad
ju
s
ted
b
ased
o
n
s
p
ec
if
ic
n
ee
d
s
an
d
co
n
te
x
ts
[
1
6
]
.
2
.
3
.
Da
t
a
clea
nin
g
Data
clea
n
in
g
is
p
er
f
o
r
m
ed
to
en
s
u
r
e
o
f
h
ig
h
q
u
ality
an
d
r
eliab
le
in
p
u
t
f
o
r
th
e
m
o
d
els.
T
h
is
s
tep
in
v
o
lv
es
th
e
h
an
d
lin
g
o
f
m
is
s
in
g
d
ata,
o
u
tlier
s
,
d
ata
er
r
o
r
s
an
d
d
ata
ty
p
es
[
1
0
]
.
Ad
d
itio
n
ally
,
f
ilter
in
g
is
ap
p
lied
to
r
em
o
v
e
co
n
s
tan
t v
ar
iab
les,
wh
ich
p
r
o
v
id
e
n
o
u
s
ef
u
l in
f
o
r
m
atio
n
a
n
d
d
u
p
licate
v
ar
iab
les,
wh
ich
in
tr
o
d
u
ce
r
ed
u
n
d
a
n
cy
.
Pro
p
er
d
ata
clea
n
i
n
g
,
in
clu
d
i
n
g
th
ese
f
ilter
in
g
s
tep
s
,
is
n
ec
e
s
s
ar
y
f
o
r
th
e
m
o
d
el
to
lear
n
f
r
o
m
ac
cu
r
ate
an
d
r
ep
r
esen
tativ
e
d
ata
[
1
]
.
2
.
4
.
F
e
a
t
ure
s
elec
t
io
n
Featu
r
e
s
elec
tio
n
is
a
cr
itical
p
r
o
ce
s
s
in
m
ac
h
in
e
lear
n
in
g
.
I
t
in
v
o
lv
es
s
elec
tin
g
a
r
elev
an
t
s
u
b
s
et
o
f
in
p
u
t
v
ar
iab
les
f
r
o
m
a
lar
g
er
d
ataset
[
1
7
]
.
T
h
e
p
r
im
ar
y
o
b
jectiv
e
o
f
f
ea
tu
r
e
s
elec
tio
n
is
t
o
id
en
tif
y
th
e
m
o
s
t
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J I
n
f
&
C
o
m
m
u
n
T
ec
h
n
o
l
I
SS
N:
2252
-
8
7
7
6
E
xp
lo
r
in
g
effec
tive
fea
tu
r
e
s
elec
tio
n
meth
o
d
s
in
lo
g
is
t
ic
r
eg
r
ess
io
n
fo
r
cred
it
…
(
B
r
ig
itta
P
erma
d
h
i
)
1617
in
f
o
r
m
ativ
e
an
d
s
ig
n
if
ican
t
f
e
atu
r
es
th
at
co
n
tr
ib
u
te
th
e
m
o
s
t
to
th
e
p
r
ed
ictiv
e
ab
ilit
y
o
f
a
m
ac
h
in
e
lear
n
in
g
m
o
d
el
[
1
8
]
.
Sev
er
al
f
ea
t
u
r
e
s
elec
tio
n
m
eth
o
d
s
ex
is
t,
in
clu
d
in
g
f
ilter
,
wr
a
p
p
er
,
an
d
em
b
e
d
d
ed
m
eth
o
d
s
[
1
9
]
.
T
h
e
f
ilter
m
eth
o
d
ev
alu
ates
e
ac
h
f
ea
tu
r
e
in
d
iv
id
u
ally
u
s
in
g
s
tatis
t
ical
tech
n
iq
u
es
to
d
eter
m
in
e
it
s
r
elev
an
ce
lev
el
to
t
h
e
tar
g
et
v
a
r
iab
le.
T
h
e
f
ilter
m
eth
o
d
h
elp
s
id
e
n
tify
a
n
d
r
em
o
v
e
co
n
s
tan
t,
d
u
p
licate
o
r
h
ig
h
l
y
co
r
r
elate
d
f
ea
tu
r
es in
th
e
d
ataset
[
1
0
]
.
T
h
e
wr
ap
p
er
m
eth
o
d
is
an
iter
ativ
e
ap
p
r
o
ac
h
th
at
s
elec
ts
an
o
p
tim
al
s
u
b
s
et
o
f
v
ar
iab
les
f
o
r
a
p
r
ed
i
ctiv
e
m
o
d
el
[
1
5
]
.
I
n
ea
c
h
iter
atio
n
,
a
m
o
d
el
is
b
u
ilt
u
s
in
g
th
e
s
elec
ted
v
ar
iab
les
an
d
ev
alu
ated
u
s
in
g
m
etr
ics
lik
e
ac
cu
r
ac
y
o
r
F1
-
s
co
r
e
[
2
0
]
.
T
h
e
p
r
o
ce
s
s
co
n
tin
u
es
u
n
til
s
to
p
p
in
g
cr
iter
ia
ar
e
m
et.
T
h
is
m
eth
o
d
ac
co
u
n
ts
f
o
r
v
ar
ia
b
le
in
ter
ac
ti
o
n
s
an
d
is
ef
f
e
ctiv
e
f
o
r
h
a
n
d
li
n
g
co
r
r
elate
d
o
r
less
im
p
ac
tf
u
l
v
ar
iab
les,
en
s
u
r
i
n
g
m
o
r
e
ac
cu
r
ate
a
n
d
o
p
tim
al
f
ea
tu
r
e
s
elec
tio
n
[
2
1
]
.
I
n
ad
d
itio
n
,
it
is
g
o
o
d
at
r
em
o
v
in
g
r
e
d
u
n
d
a
n
t
o
r
wea
k
ly
co
r
r
elate
d
f
ea
tu
r
es.
T
h
is
h
ap
p
en
s
b
ec
au
s
e,
b
y
test
in
g
th
e
ef
f
ec
ts
o
n
a
m
o
d
el'
s
p
er
f
o
r
m
a
n
ce
,
th
is
m
eth
o
d
ca
n
cu
t
o
u
t
all
th
e
v
ar
iab
les
wh
i
ch
d
o
n
o
t
ad
d
a
n
y
th
in
g
to
t
h
e
m
o
d
el'
s
p
r
ed
ictiv
e
ca
p
ab
il
ity
.
Ho
wev
er
,
it
is
tim
e
-
co
n
s
u
m
in
g
f
o
r
lar
g
e
-
s
ca
le
d
ata
d
u
e
to
its
co
m
p
u
tati
o
n
al
in
ef
f
icien
cy
.
W
r
ap
p
e
r
is
g
r
ee
d
y
an
d
co
u
ld
ea
s
ily
g
et
s
tu
ck
in
a
lo
ca
l
o
p
ti
m
u
m
.
T
h
e
f
ea
t
u
r
e
s
elec
tio
n
p
r
o
ce
s
s
wo
u
ld
th
en
b
e
to
o
d
ep
e
n
d
en
t
o
n
v
alid
atio
n
p
er
f
o
r
m
an
ce
,
d
is
ca
r
d
u
s
ef
u
l
f
ea
tu
r
es
to
o
a
g
g
r
ess
iv
ely
,
an
d
m
ay
ca
u
s
e
u
n
d
er
f
itti
n
g
.
I
n
co
n
tr
ast,
th
e
o
th
er
s
h
o
r
tco
m
in
g
o
f
th
e
wr
ap
p
e
r
is
it
m
ig
h
t
ca
u
s
e
o
v
er
f
it
tin
g
.
B
y
r
ep
ea
te
d
ly
o
p
tim
iz
in
g
f
o
r
v
alid
atio
n
p
er
f
o
r
m
an
ce
,
it
ca
n
f
it
to
o
clo
s
ely
to
th
e
tr
ain
i
n
g
d
a
ta,
ca
p
tu
r
in
g
n
o
is
e
an
d
u
n
n
ec
ess
ar
y
f
ea
tu
r
es,
wh
ich
d
im
in
is
h
es th
e
m
o
d
el’
s
ab
ilit
y
to
g
en
er
alize
t
o
u
n
s
ee
n
d
ata.
T
h
e
em
b
ed
d
ed
m
eth
o
d
in
teg
r
ates
f
ea
tu
r
e
s
elec
tio
n
in
t
o
th
e
tr
ain
in
g
p
r
o
ce
s
s
,
allo
win
g
th
e
alg
o
r
ith
m
to
au
to
m
atica
lly
s
elec
t
th
e
m
o
s
t
im
p
o
r
tan
t
f
ea
tu
r
es
[
2
2
]
.
Sig
n
if
ican
t
f
ea
tu
r
es
ar
e
g
iv
en
h
ig
h
er
weig
h
ts
,
wh
ile
less
im
p
o
r
tan
t
o
n
es
ar
e
d
is
r
eg
ar
d
ed
[
2
3
]
.
Ho
wev
er
,
t
h
is
m
eth
o
d
m
ay
s
tr
u
g
g
le
with
h
ig
h
ly
co
r
r
elate
d
f
ea
tu
r
es
an
d
m
ay
n
o
t
alwa
y
s
p
r
o
v
id
e
p
r
ec
is
e
im
p
o
r
tan
ce
r
an
k
i
n
g
s
[
7
]
.
A
m
ajo
r
wea
k
n
ess
is
o
v
er
-
p
e
n
aliza
tio
n
th
r
o
u
g
h
L
ass
o
r
eg
u
lar
izatio
n
:
it
f
o
r
ce
s
to
o
m
an
y
f
ea
tu
r
es
to
b
e
ze
r
o
,
d
esp
ite
p
o
ten
tially
h
av
in
g
co
m
p
lem
en
tar
y
v
alu
e
to
o
th
er
f
ea
tu
r
es.
T
h
is
m
ay
lea
d
to
u
n
d
er
f
itti
n
g
as
L
ass
o
d
is
c
ar
d
s
im
p
o
r
tan
t
f
ea
tu
r
es.
I
n
ad
d
itio
n
,
its
em
p
h
asis
o
n
e
v
alu
atin
g
f
ea
tu
r
es
i
n
d
iv
i
d
u
ally
m
ea
n
s
th
at
f
ea
tu
r
e
in
ter
ac
tio
n
s
o
r
co
m
b
in
ed
ef
f
ec
ts
o
f
f
ea
t
u
r
es
ar
e
o
v
er
lo
o
k
ed
,
d
esp
ite
th
eir
p
o
te
n
tial
to
im
p
r
o
v
e
p
r
ed
ictiv
e
ac
c
u
r
ac
y
.
T
h
is
lim
itatio
n
b
ec
o
m
e
s
m
o
r
e
p
r
o
n
o
u
n
ce
d
wh
en
f
ea
tu
r
es
ar
e
h
i
g
h
ly
c
o
r
r
elate
d
,
s
in
ce
it
ty
p
ically
r
et
ain
s
o
n
ly
a
s
in
g
le
v
ar
ia
b
le
f
r
o
m
th
e
g
r
o
u
p
an
d
r
em
o
v
es
f
ea
t
u
r
es
th
at
s
till
ca
r
r
y
in
f
o
r
m
ativ
e
s
ig
n
als,
lead
in
g
to
u
n
d
e
r
f
itti
n
g
.
Fu
r
th
er
m
o
r
e
,
L
ass
o
s
tr
u
g
g
les
to
ca
p
tu
r
e
n
o
n
-
lin
ea
r
r
elatio
n
s
h
i
p
s
b
etwe
en
f
ea
tu
r
es
an
d
th
e
tar
g
et
v
ar
ia
b
le,
lim
itin
g
its
ca
p
ab
ilit
y
to
m
o
d
el
co
m
p
lex
p
atter
n
s
an
d
i
n
cr
ea
s
in
g
th
e
r
is
k
o
f
u
n
d
er
f
itti
n
g
in
s
u
ch
s
ce
n
ar
io
s
.
T
h
e
h
y
b
r
id
m
eth
o
d
in
t
h
is
s
tu
d
y
co
m
b
i
n
es
em
b
e
d
d
ed
an
d
wr
ap
p
e
r
f
ea
t
u
r
e
s
elec
tio
n
tec
h
n
iq
u
es
t
o
im
p
r
o
v
e
b
o
th
m
o
d
el
p
er
f
o
r
m
a
n
ce
an
d
g
en
er
aliza
tio
n
.
W
h
ile
em
b
ed
d
ed
m
eth
o
d
s
s
u
ch
as
L
ass
o
ar
e
ef
f
ec
tiv
e
in
r
ed
u
cin
g
d
im
en
s
io
n
ality
th
r
o
u
g
h
r
e
g
u
lar
izatio
n
,
wr
a
p
p
er
m
eth
o
d
s
s
u
ch
as
Seq
u
en
tial
Fo
r
war
d
Selectio
n
(
SF
S)
ar
e
ca
p
ab
le
o
f
ca
p
tu
r
i
n
g
in
ter
ac
tio
n
s
b
etwe
en
f
ea
tu
r
es.
B
y
in
teg
r
atin
g
th
ese
ap
p
r
o
ac
h
es,
th
e
h
y
b
r
id
m
eth
o
d
aim
s
to
b
alan
ce
s
tatis
tical
r
elev
an
ce
an
d
p
r
e
d
ictiv
e
p
er
f
o
r
m
a
n
ce
[
1
0
]
.
I
n
th
is
im
p
lem
en
tatio
n
,
th
e
h
y
b
r
id
ap
p
r
o
ac
h
is
co
n
d
u
cte
d
in
two
s
tag
es.
First,
L
1
-
b
ased
L
a
s
s
o
r
eg
u
lar
izatio
n
is
ap
p
lied
to
t
h
e
d
ataset
to
elim
in
ate
less
r
elev
an
t
f
ea
tu
r
es
b
y
s
h
r
i
n
k
in
g
th
eir
co
e
f
f
icien
ts
to
war
d
ze
r
o
.
T
h
e
r
e
g
u
lar
izati
o
n
s
tr
en
g
t
h
is
co
n
t
r
o
lled
t
o
b
alan
ce
b
etwe
en
f
ea
tu
r
e
s
p
a
r
s
ity
an
d
in
f
o
r
m
atio
n
r
eten
tio
n
.
Seco
n
d
,
Seq
u
en
tial
Fo
r
war
d
Selectio
n
(
SF
S)
is
ap
p
lied
to
th
e
r
ed
u
ce
d
f
ea
tu
r
e
s
et.
I
n
th
is
s
tag
e,
f
ea
tu
r
es
ar
e
ad
d
ed
iter
ativ
el
y
b
ased
o
n
t
h
eir
co
n
tr
ib
u
tio
n
to
m
o
d
el
p
e
r
f
o
r
m
an
ce
,
u
s
in
g
L
o
g
is
tic
R
eg
r
ess
io
n
as
th
e
b
ase
esti
m
ato
r
.
T
h
e
s
elec
tio
n
p
r
o
ce
s
s
co
n
tin
u
es
u
n
til
n
o
s
ig
n
if
ican
t
i
m
p
r
o
v
em
e
n
t
in
ev
alu
atio
n
m
etr
ics
is
o
b
s
er
v
ed
.
T
h
is
two
-
s
tag
e
p
r
o
ce
s
s
en
s
u
r
es
th
at
th
e
s
elec
te
d
f
ea
tu
r
es
ar
e
b
o
th
s
tatis
tical
ly
m
ea
n
in
g
f
u
l
an
d
em
p
ir
ically
ef
f
ec
tiv
e
f
o
r
p
r
ed
i
ctio
n
[
2
3
]
.
T
h
is
s
tu
d
y
ap
p
lies
f
o
u
r
f
ea
t
u
r
e
s
elec
tio
n
m
eth
o
d
s
:
wr
ap
p
er
,
em
b
ed
d
ed
,
h
y
b
r
id
,
an
d
n
o
f
ea
tu
r
e
s
elec
tio
n
.
T
h
e
wr
a
p
p
er
m
eth
o
d
u
s
es
SF
S
to
iter
ativ
ely
s
elec
t
o
p
tim
al
f
ea
tu
r
es,
wh
ile
th
e
em
b
ed
d
ed
m
eth
o
d
ap
p
lies
L
ass
o
(
L
1
)
r
e
g
u
lar
iza
tio
n
to
elim
in
ate
less
im
p
o
r
ta
n
t
f
ea
tu
r
es.
T
h
e
h
y
b
r
id
m
et
h
o
d
,
as
th
e
p
r
o
p
o
s
ed
ap
p
r
o
ac
h
,
is
co
m
p
ar
e
d
ag
ain
s
t
th
ese
co
n
v
en
tio
n
al
m
eth
o
d
s
to
ass
es
s
p
er
f
o
r
m
an
ce
.
E
ac
h
s
elec
ted
f
ea
tu
r
e
s
u
b
s
et
is
te
s
ted
u
s
in
g
lo
g
is
tic
r
eg
r
ess
io
n
,
an
d
th
e
r
esu
lts
ar
e
ev
alu
ated
u
s
in
g
a
co
n
f
u
s
io
n
m
atr
ix
,
ac
cu
r
ac
y
,
p
r
ec
is
io
n
,
r
ec
all,
an
d
F1
-
s
co
r
e
.
2
.
5
.
Da
t
a
s
pli
t
t
ing
T
h
e
clea
n
ed
d
ataset
is
s
p
lit
i
n
to
tr
ain
in
g
an
d
test
in
g
s
u
b
s
ets
u
s
in
g
a
Ho
ld
o
u
t
v
alid
atio
n
ap
p
r
o
ac
h
with
a
7
0
:3
0
r
ati
o
.
T
o
en
s
u
r
e
th
at
th
e
class
d
is
tr
ib
u
tio
n
r
em
ain
s
co
n
s
is
ten
t
ac
r
o
s
s
b
o
t
h
s
u
b
s
ets,
s
tr
atif
ied
s
am
p
lin
g
is
ap
p
lied
d
u
r
i
n
g
th
e
s
p
litt
in
g
p
r
o
ce
s
s
.
T
h
is
ap
p
r
o
a
ch
p
r
eser
v
es
th
e
p
r
o
p
o
r
tio
n
o
f
“g
o
o
d
”
a
n
d
“b
a
d
”
cr
ed
it c
lass
es,
th
er
eb
y
r
ed
u
ci
n
g
s
am
p
lin
g
b
ias an
d
en
s
u
r
in
g
a
f
air
ev
alu
atio
n
o
f
m
o
d
el
p
er
f
o
r
m
an
ce
o
n
u
n
s
ee
n
d
ata.
T
h
e
tr
ain
in
g
s
u
b
s
et
is
u
s
ed
to
b
u
ild
a
n
d
o
p
tim
ize
th
e
m
o
d
el,
wh
ile
th
e
test
in
g
s
u
b
s
et
ev
alu
ates
its
p
er
f
o
r
m
an
ce
o
n
u
n
s
ee
n
d
ata,
a
s
s
es
s
in
g
h
o
w
well
th
e
m
o
d
el
g
en
er
alize
s
to
r
ea
l
-
wo
r
ld
s
itu
ati
o
n
s
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8
7
7
6
I
n
t J I
n
f
&
C
o
m
m
u
n
T
ec
h
n
o
l
,
Vo
l.
1
5
,
No
.
4
,
D
e
c
em
b
er
20
2
6
:
1
6
1
4
-
1
6
2
3
1618
2
.
6
.
M
o
del c
re
a
t
io
n
T
h
e
lo
g
is
tic
r
eg
r
ess
io
n
m
o
d
e
l
is
b
u
ilt
u
s
in
g
th
e
tr
ain
in
g
d
ata
to
p
r
ed
ict
a
b
i
n
ar
y
o
u
tc
o
m
e
[
2
0
]
.
L
o
g
is
tic
r
eg
r
ess
io
n
m
o
d
els
th
e
r
elatio
n
s
h
ip
b
etwe
en
i
n
p
u
t
v
ar
iab
les
(
)
a
n
d
th
e
p
r
o
b
ab
il
ity
o
f
an
ev
en
t
(
)
o
cc
u
r
r
in
g
,
p
r
o
d
u
cin
g
b
in
ar
y
o
u
tp
u
ts
(
0
o
r
1
)
[
1
5
]
.
T
h
is
is
ac
h
iev
ed
th
r
o
u
g
h
th
e
s
ig
m
o
i
d
f
u
n
ctio
n
,
wh
ich
m
ap
s
a
lin
ea
r
co
m
b
in
atio
n
o
f
th
e
i
n
p
u
ts
an
d
th
eir
r
esp
ec
tiv
e
weig
h
ts
to
a
p
r
o
b
ab
ilit
y
v
alu
e)
[
1
5
]
.
T
h
e
r
esu
ltin
g
p
r
o
b
a
b
ilit
y
is
th
en
s
u
b
jecte
d
to
a
th
r
esh
o
ld
to
d
eter
m
in
e
th
e
class
if
icatio
n
o
u
tp
u
t,
with
ad
ju
s
tm
en
ts
m
ad
e
to
b
alan
ce
b
etwe
en
ac
c
u
r
ac
y
a
n
d
m
in
im
izin
g
f
alse n
e
g
ativ
es [
1
6
]
.
T
r
ain
in
g
an
d
test
in
g
p
h
ases
ar
e
b
o
th
in
v
o
lv
e
d
in
th
e
lo
g
is
tic
r
eg
r
ess
io
n
p
r
o
ce
s
s
.
T
h
e
m
o
d
el
is
tr
ain
ed
o
n
lab
eled
d
ata,
a
d
ju
s
tin
g
its
weig
h
ts
to
in
cr
ea
s
e
th
e
ac
cu
r
a
cy
o
f
its
p
r
ed
icted
o
u
tco
m
es.
T
esti
n
g
,
o
n
th
e
o
t
h
er
h
an
d
,
is
p
er
f
o
r
m
e
d
b
y
ap
p
l
y
in
g
th
e
lear
n
ed
co
e
f
f
icien
ts
to
n
ew
d
ata.
T
h
is
allo
ws
th
e
r
es
ea
r
ch
er
to
g
en
er
ate
ac
cu
r
ate
an
d
in
ter
p
r
etab
le
p
r
ed
ictio
n
s
th
at
ca
n
b
e
u
s
ed
in
a
wid
e
r
an
g
e
o
f
ap
p
licatio
n
s
b
y
esti
m
atin
g
th
e
p
r
o
b
a
b
ilit
y
o
f
a
s
p
ec
if
ic
o
u
tco
m
e.
Du
r
in
g
m
o
d
el
cr
ea
tio
n
,
it
is
c
r
u
cial
to
b
alan
ce
o
v
er
f
itti
n
g
a
n
d
u
n
d
e
r
f
itti
n
g
[
2
4
]
.
Ov
er
f
itti
n
g
o
cc
u
r
s
wh
en
a
m
o
d
el
is
to
o
co
m
p
le
x
an
d
"m
em
o
r
izes"
th
e
tr
ain
in
g
d
ata,
ca
p
tu
r
in
g
n
o
is
e
an
d
o
u
tlier
s
in
s
tead
o
f
g
en
er
al
p
atter
n
s
,
lead
in
g
to
lo
w
tr
ain
in
g
er
r
o
r
b
u
t
h
i
g
h
te
s
tin
g
er
r
o
r
.
Ad
d
r
ess
in
g
o
v
er
f
itti
n
g
m
ay
in
v
o
lv
e
ad
d
in
g
d
ata,
r
eg
u
lar
izatio
n
o
r
r
ed
u
cin
g
m
o
d
el
c
o
m
p
lex
it
y
[
7
]
.
I
n
co
n
tr
ast,
u
n
d
er
f
itti
n
g
o
cc
u
r
s
wh
en
th
e
m
o
d
el
i
s
n
o
t
s
o
p
h
is
ticated
en
o
u
g
h
to
id
en
tif
y
u
n
d
er
ly
in
g
p
atter
n
s
,
lead
in
g
to
h
ig
h
tr
ain
in
g
an
d
test
in
g
er
r
o
r
[
5
]
.
Po
s
s
ib
le
ap
p
r
o
ac
h
es
to
ad
d
r
ess
in
g
u
n
d
er
f
itti
n
g
in
clu
d
e
in
cr
ea
s
in
g
m
o
d
el
co
m
p
le
x
ity
o
r
in
co
r
p
o
r
atin
g
ad
d
itio
n
al
r
e
lev
an
t
f
ea
tu
r
es.
Ma
in
tain
in
g
a
b
ala
n
ce
b
etwe
en
u
n
d
er
f
itti
n
g
an
d
o
v
er
f
itti
n
g
is
ess
en
tial
f
o
r
r
eliab
le
m
o
d
el
p
er
f
o
r
m
a
n
ce
.
C
o
m
m
o
n
s
tr
ateg
i
es
u
s
ed
to
ac
h
ie
v
e
th
is
b
alan
ce
in
clu
d
e
r
eg
u
lar
iza
tio
n
an
d
f
ea
tu
r
e
s
elec
tio
n
.
Af
ter
th
e
lo
g
is
tic
r
eg
r
ess
io
n
m
o
d
el
is
b
u
ilt
u
s
in
g
th
e
tr
ain
in
g
d
ata
a
n
d
p
ass
ed
th
e
s
atis
f
ac
to
r
y
ev
alu
atio
n
,
we
ca
n
u
s
e
it
to
m
ak
e
p
r
ed
ictio
n
s
o
n
n
ew
d
ata.
B
y
in
p
u
ttin
g
th
e
v
alu
es
o
f
th
e
in
p
u
t
v
ar
iab
les
in
to
th
e
m
o
d
el,
we
ca
n
o
b
tain
th
e
p
r
ed
icted
v
alu
es o
f
th
e
o
u
tp
u
t
v
ar
iab
le
o
r
th
e
p
r
o
b
ab
ilit
ies.
T
o
en
s
u
r
e
r
eliab
le
m
o
d
el
p
er
f
o
r
m
an
ce
,
s
ev
er
al
im
p
le
m
en
tatio
n
c
o
n
s
id
er
atio
n
s
ar
e
ap
p
lied
.
R
eg
u
lar
izatio
n
is
u
s
ed
in
L
o
g
is
tic
R
eg
r
ess
io
n
to
p
r
ev
en
t
o
v
er
f
itti
n
g
.
C
lass
im
b
alan
ce
is
h
an
d
led
th
r
o
u
g
h
s
tr
atif
ied
s
am
p
lin
g
d
u
r
i
n
g
d
ata
s
p
litt
in
g
to
p
r
eser
v
e
class
d
is
tr
ib
u
tio
n
.
I
n
ad
d
itio
n
,
p
r
e
p
r
o
ce
s
s
in
g
s
tep
s
in
clu
d
e
th
e
r
em
o
v
al
o
f
c
o
n
s
tan
t a
n
d
h
i
g
h
ly
co
r
r
elate
d
f
ea
tu
r
es to
r
ed
u
ce
n
o
is
e
an
d
r
ed
u
n
d
a
n
cy
in
t
h
e
d
ataset.
2
.
7
.
M
o
del e
v
a
lua
t
i
o
n
T
h
e
ev
alu
atio
n
m
eth
o
d
in
v
o
l
v
es
u
s
in
g
a
co
n
f
u
s
io
n
m
atr
ix
t
o
test
m
o
d
el
p
er
f
o
r
m
a
n
ce
an
d
ac
cu
r
ac
y
o
n
th
e
test
d
ata
[
1
9
]
.
T
h
is
m
atr
ix
is
a
s
u
m
m
a
r
y
tab
le
t
h
at
ca
p
tu
r
es
t
h
e
m
o
d
el'
s
co
r
r
ec
t
an
d
in
co
r
r
ec
t
p
r
ed
ictio
n
s
b
y
co
m
p
ar
in
g
p
r
ed
icted
class
lab
els
with
ac
t
u
al
class
lab
el
s
.
T
h
e
co
n
f
u
s
io
n
m
atr
ix
h
as
f
o
u
r
ess
en
tial
co
m
p
o
n
en
ts
:
tr
u
e
p
o
s
itiv
e
(
T
P),
tr
u
e
n
eg
ativ
e
(
T
N)
,
f
als
e
p
o
s
itiv
e
(
FP
)
an
d
f
alse
n
eg
ativ
e
(
FN)
[
2
5
]
.
T
r
u
e
p
o
s
itiv
es
r
ep
r
esen
t
th
e
p
o
s
itiv
e
ca
s
es
th
at
ar
e
co
r
r
e
ctly
class
if
ied
an
d
T
r
u
e
Neg
ativ
es
is
th
e
ca
s
es
co
r
r
ec
tly
id
en
tifie
d
as
n
eg
ati
v
e.
Fals
e
Po
s
itiv
es
r
ep
r
esen
t
in
s
tan
ce
s
in
co
r
r
ec
tly
class
if
ie
d
as
p
o
s
itiv
e
,
ev
en
th
o
u
g
h
th
e
y
ar
e
n
eg
ati
v
e
an
d
Fals
e
Neg
ativ
es
d
en
o
te
ca
s
es
in
co
r
r
ec
tly
class
if
ied
as
n
eg
ativ
e
d
esp
ite
b
ei
n
g
p
o
s
itiv
e
[
2
4
]
.
T
h
is
a
p
p
r
o
ac
h
co
n
s
titu
tes
th
e
p
r
im
ar
y
e
v
al
u
atio
n
m
eth
o
d
u
s
ed
in
th
is
s
tu
d
y
,
p
r
o
v
i
d
in
g
a
s
tan
d
ar
d
ized
f
r
am
ewo
r
k
f
o
r
a
s
s
es
s
in
g
cl
ass
if
icatio
n
p
er
f
o
r
m
an
ce
.
Usi
n
g
th
ese
co
m
p
o
n
e
n
ts
o
f
t
h
e
co
n
f
u
s
io
n
m
atr
ix
,
we
d
er
i
v
e
k
ey
m
etr
ics th
at
p
r
o
v
i
d
e
in
s
ig
h
ts
in
to
th
e
m
o
d
el’
s
p
er
f
o
r
m
a
n
ce
s
u
ch
as:
Acc
u
r
ac
y
is
th
e
o
v
er
all
co
r
r
ec
t
n
ess
r
ate,
ca
lcu
lated
b
y
(
1
)
.
=
+
+
+
+
(
1
)
Pre
cisi
o
n
is
th
e
p
r
o
p
o
r
tio
n
o
f
t
r
u
e
p
o
s
itiv
e
p
r
ed
ictio
n
s
am
o
n
g
p
o
s
itiv
e
p
r
e
d
ictio
n
s
,
ca
lcu
lat
ed
b
y
(
2
)
.
=
+
(
2
)
R
ec
all
o
r
Sen
s
itiv
ity
i
s
th
e
p
r
o
p
o
r
tio
n
o
f
tr
u
e
p
o
s
itiv
e
p
r
ed
ictio
n
s
am
o
n
g
ac
tu
al
p
o
s
itiv
e
in
s
tan
ce
s
,
ca
lcu
lated
b
y
(
3
)
.
=
+
(
3
)
F1
-
s
co
r
e
r
ep
r
esen
ts
th
e
b
alan
c
e
b
etwe
en
p
r
ec
is
io
n
a
n
d
r
ec
all
,
ca
lcu
lated
b
y
(
4
)
.
1
−
=
2
∗
∗
+
(
4
)
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J I
n
f
&
C
o
m
m
u
n
T
ec
h
n
o
l
I
SS
N:
2252
-
8
7
7
6
E
xp
lo
r
in
g
effec
tive
fea
tu
r
e
s
elec
tio
n
meth
o
d
s
in
lo
g
is
t
ic
r
eg
r
ess
io
n
fo
r
cred
it
…
(
B
r
ig
itta
P
erma
d
h
i
)
1619
C
o
m
p
ar
in
g
th
e
m
etr
ic
v
alu
e
s
f
r
o
m
th
e
s
elec
ted
f
ea
tu
r
e
-
s
elec
tio
n
m
eth
o
d
with
th
o
s
e
f
r
o
m
th
e
co
n
v
en
tio
n
al
m
o
d
els
will
h
elp
clar
if
y
th
eir
r
elativ
e
p
e
r
f
o
r
m
a
n
ce
.
T
h
is
ev
alu
atio
n
s
u
p
p
o
r
ts
co
n
s
is
ten
t
co
m
p
ar
is
o
n
ac
r
o
s
s
d
if
f
er
en
t
m
o
d
el
co
n
f
i
g
u
r
atio
n
s
an
d
ass
is
ts
in
id
en
tif
y
in
g
th
e
m
o
s
t
s
u
itab
le
m
o
d
el
f
o
r
th
e
task
.
I
t a
ls
o
h
elp
s
in
d
icate
ar
ea
s
th
at
m
ay
r
eq
u
ir
e
f
u
r
th
er
im
p
r
o
v
em
en
t
o
r
clo
s
er
atten
tio
n
.
2
.
8
.
M
o
del v
a
lid
a
t
io
n
Af
ter
b
u
ild
in
g
th
e
m
o
d
el
an
d
s
elec
tin
g
th
e
b
est
s
u
b
s
et,
it
c
an
b
e
v
alid
ated
o
n
n
ew
d
ata
to
co
n
f
ir
m
it’
s
o
p
tim
al.
Mo
d
el
v
alid
atio
n
ass
ess
e
s
th
e
m
o
d
el’
s
g
en
er
al
izatio
n
ab
ilit
y
a
n
d
ac
c
u
r
ac
y
with
n
ew
in
s
tan
ce
s
,
co
m
p
ar
in
g
p
r
e
d
ictio
n
s
to
ac
tu
al
o
u
tco
m
es
[
1
3
]
.
Key
ev
alu
a
tio
n
m
etr
ics
lik
e
ac
cu
r
ac
y
,
p
r
ec
is
io
n
,
r
ec
all
an
d
F1
-
s
co
r
e
h
elp
d
ete
r
m
in
e
th
e
m
o
d
el’
s
r
eliab
ilit
y
.
I
f
t
h
e
m
o
d
el
p
er
f
o
r
m
s
s
atis
f
ac
to
r
ily
,
it
is
d
ee
m
ed
s
u
itab
le
f
o
r
r
ea
l w
o
r
ld
ap
p
licatio
n
s
[
2
4
]
.
I
n
ad
d
itio
n
to
th
e
Ho
ld
o
u
t
v
alid
atio
n
ap
p
r
o
ac
h
,
5
-
f
o
ld
cr
o
s
s
-
v
alid
atio
n
is
co
n
d
u
cted
to
f
u
r
th
er
ev
alu
ate
m
o
d
el
r
o
b
u
s
tn
ess
.
T
h
e
d
ataset
is
d
iv
id
ed
in
to
f
iv
e
s
u
b
s
ets,
an
d
th
e
m
o
d
el
is
tr
ain
ed
an
d
ev
alu
ate
d
iter
ativ
ely
ac
r
o
s
s
d
if
f
er
e
n
t
f
o
ld
s
.
T
h
e
av
er
a
g
e
ac
cu
r
ac
y
a
cr
o
s
s
f
o
ld
s
is
u
s
ed
to
p
r
o
v
i
d
e
a
m
o
r
e
r
eliab
l
e
esti
m
ate
o
f
m
o
d
el
p
e
r
f
o
r
m
an
c
e
an
d
to
r
ed
u
ce
s
en
s
itiv
ity
to
d
ata
p
ar
titi
o
n
in
g
.
2
.
9
.
Resul
t
a
na
ly
s
is
Af
ter
th
e
ev
al
u
atio
n
p
h
ase,
th
is
s
tu
d
y
p
er
f
o
r
m
s
a
n
an
al
y
s
is
in
a
s
tr
u
ctu
r
ed
co
m
p
ar
i
s
o
n
o
f
th
e
o
u
tco
m
es
f
r
o
m
th
e
ap
p
lied
f
ea
tu
r
e
s
elec
tio
n
m
eth
o
d
s
.
Acc
u
r
ac
y
,
p
r
ec
is
io
n
,
r
ec
all
an
d
F1
-
s
co
r
e
in
th
e
tr
ain
in
g
an
d
test
in
g
s
ets
ar
e
an
aly
ze
d
,
in
o
r
d
er
to
ev
alu
ate
th
e
p
r
e
d
ictiv
e
q
u
ality
.
T
est
-
tr
ain
d
if
f
er
en
ce
s
ar
e
f
u
r
th
er
ca
lcu
lated
an
d
c
o
m
p
ar
e
d
in
o
r
d
er
t
o
g
et
a
s
en
s
e
f
o
r
g
en
er
aliza
tio
n
,
i
n
clu
d
in
g
o
v
er
-
an
d
u
n
d
er
f
itti
n
g
.
Misclas
s
if
icatio
n
co
u
n
ts
ar
e
f
u
r
th
er
co
m
p
ar
ed
to
aid
th
e
in
ter
p
r
etatio
n
o
f
r
eliab
ilit
y
u
n
d
er
class
im
b
alan
ce
.
Fin
ally
,
s
elec
ted
f
ea
tu
r
es
ar
e
b
r
ief
ly
r
ev
iewe
d
f
o
r
r
ele
v
an
ce
an
d
d
iv
er
s
ity
.
T
h
e
m
o
d
el
with
th
e
b
est
p
er
f
o
r
m
in
g
g
en
er
aliza
tio
n
is
i
d
en
tifie
d
as m
o
s
t su
itab
le
b
ase
d
o
n
th
is
an
al
y
s
is
.
3.
RE
SU
L
T
S AN
D
D
I
SCU
SS
I
O
N
I
n
th
is
p
ap
er
we
ar
e
ev
alu
atin
g
an
d
co
m
p
a
r
in
g
th
r
ee
o
f
f
ea
tu
r
e
s
elec
tio
n
m
eth
o
d
s
,
wr
ap
p
er
s
,
em
b
ed
d
e
d
a
n
d
h
y
b
r
id
m
eth
o
d
s
u
n
d
er
o
n
e
l
o
g
is
tic
r
eg
r
ess
io
n
m
o
d
el,
t
o
a
p
p
ly
in
cr
ed
it
s
co
r
in
g
.
T
h
e
p
r
o
ject
aim
s
to
cr
ea
te
a
tem
p
late
f
o
r
p
er
f
o
r
m
in
g
th
e
f
ea
t
u
r
e
s
elec
tio
n
ef
f
ec
tiv
ely
,
f
o
r
n
u
m
e
r
ical
d
ata,
to
b
u
ild
co
n
s
is
ten
t p
r
ed
ictiv
e
m
o
d
els with
n
o
o
r
v
er
y
litt
le
o
v
er
f
itti
n
g
[
1
7
]
.
T
h
e
d
ataset
is
a
cr
ed
it
s
co
r
in
g
d
ataset
f
r
o
m
a
b
an
k
in
g
in
f
o
r
m
atio
n
wh
ich
co
n
s
is
ts
o
f
6
6
c
o
lu
m
n
s
a
n
d
1
5
3
,
5
8
8
r
o
ws,
4
0
o
f
t
h
e
f
ea
t
u
r
es a
r
e
co
n
s
tan
t
an
d
1
1
ar
e
h
ig
h
ly
co
r
r
elate
d
f
ea
tu
r
es r
em
o
v
ed
i
n
th
e
p
r
ep
r
o
ce
s
s
in
g
p
h
ase.
T
h
e
tar
g
et
v
ar
ia
b
le
“Ku
alitas” i
s
co
n
v
er
ted
to
two
-
class
class
if
icat
io
n
with
"g
o
o
d
"
(
1
,
2
)
an
d
"
b
ad
"
(
3
,
4
,
5
)
as a
p
p
r
o
v
al
a
n
d
d
en
ial
r
es
p
ec
tiv
ely
.
T
h
e
r
esear
ch
em
p
lo
y
s
th
r
ee
f
e
atu
r
e
s
elec
tio
n
m
o
d
el
wr
a
p
p
er
s
,
em
b
ed
d
ed
a
n
d
h
y
b
r
id
to
id
en
tify
k
ey
v
ar
iab
les
f
o
r
ac
c
u
r
ate
o
u
tco
m
e
p
r
e
d
ictio
n
[
1
8
]
.
T
h
e
wr
ap
p
er
m
eth
o
d
test
s
v
ar
io
u
s
f
ea
tu
r
e
s
u
b
s
ets
b
a
s
ed
o
n
p
er
f
o
r
m
an
ce
m
etr
ics,
o
p
tim
iz
in
g
f
o
r
th
e
b
est
co
m
b
in
atio
n
[
2
5
]
.
T
h
e
em
b
ed
d
e
d
m
eth
o
d
in
teg
r
ates
f
e
atu
r
e
s
elec
tio
n
d
ir
ec
tly
in
to
t
h
e
m
o
d
el
tr
ain
in
g
,
en
h
an
cin
g
ef
f
i
cien
cy
,
th
o
u
g
h
its
ef
f
ec
tiv
en
ess
ca
n
v
ar
y
ac
r
o
s
s
alg
o
r
ith
m
s
[
2
1
]
.
T
h
e
h
y
b
r
id
ap
p
r
o
ac
h
co
m
b
in
es
f
ilter
,
wr
ap
p
er
an
d
em
b
ed
d
e
d
m
e
th
o
d
s
,
lev
er
ag
in
g
th
eir
s
tr
en
g
th
s
to
cr
ea
te
a
co
m
p
r
eh
en
s
iv
e
an
d
b
alan
ce
d
s
elec
tio
n
p
r
o
ce
s
s
th
at
o
p
tim
izes
m
o
d
el
p
er
f
o
r
m
a
n
ce
[
2
6
]
.
E
ac
h
ap
p
r
o
ac
h
p
lay
s
a
r
o
le
in
co
n
tr
o
llin
g
o
v
er
f
itti
n
g
,
with
wr
ap
p
er
m
eth
o
d
s
f
in
e
-
tu
n
in
g
p
e
r
f
o
r
m
a
n
ce
th
r
o
u
g
h
ev
alu
atio
n
-
b
ased
s
elec
tio
n
,
an
d
em
b
ed
d
e
d
m
eth
o
d
s
ap
p
l
y
in
g
r
eg
u
lar
izatio
n
to
r
e
d
u
ce
v
ar
ian
ce
.
T
h
e
h
y
b
r
id
m
o
d
el
in
teg
r
ates th
ese
s
tr
en
g
t
h
s
to
en
h
an
ce
b
o
th
p
r
e
d
ictio
n
ac
cu
r
ac
y
an
d
g
e
n
er
aliza
tio
n
.
T
h
e
o
r
ig
in
al
d
ataset
is
s
p
lit
i
n
to
a
tr
ain
in
g
d
ataset
an
d
a
test
in
g
d
ataset,
with
th
e
f
o
r
m
er
u
s
ed
f
o
r
lear
n
in
g
th
e
m
o
d
el
p
ar
a
m
eter
s
an
d
th
e
latter
u
s
ed
f
o
r
e
v
alu
atin
g
p
r
e
d
ictiv
e
ac
cu
r
ac
y
.
I
n
t
h
is
s
tu
d
y
,
we
u
s
e
a
7
0
/3
0
r
atio
o
f
tr
ain
d
ata
to
test
d
ata
is
u
s
ed
.
T
h
is
allo
ws
th
e
m
o
d
el
to
lear
n
f
r
o
m
a
lar
g
e
en
o
u
g
h
s
am
p
le
o
f
th
e
d
ata
an
d
to
test
its
p
er
f
o
r
m
an
c
e
o
n
u
n
s
ee
n
d
ata.
T
h
e
ch
o
ice
o
f
a
lo
g
is
tic
r
eg
r
ess
io
n
m
o
d
el
was
f
o
r
th
e
ea
s
e
o
f
u
s
e
an
d
its
r
ea
d
ab
ilit
ies
o
f
i
n
ter
p
r
etatio
n
w
h
ich
ar
e
two
k
ey
f
ea
tu
r
es
in
a
cr
e
d
it
s
co
r
in
g
m
o
d
el.
L
o
g
is
tic
r
eg
r
ess
io
n
is
a
lin
ea
r
m
o
d
el
t
h
at
u
s
es
th
e
lo
g
is
ti
c
f
u
n
ctio
n
to
m
o
d
el
th
e
p
r
o
b
a
b
ilit
y
o
f
a
n
ev
en
t,
m
ak
in
g
it
a
n
atu
r
al
ch
o
ice
f
o
r
b
in
ar
y
class
if
icatio
n
task
s
s
u
ch
as
cr
ed
it
s
co
r
in
g
,
wh
er
e
th
e
o
u
tco
m
e
is
eith
er
"d
ef
a
u
lt"
o
r
"n
o
n
-
d
ef
au
lt".
T
h
e
ac
cu
r
ac
y
o
f
th
e
m
o
d
el
is
m
ea
s
u
r
ed
b
y
th
e
p
er
ce
n
tag
e
o
f
c
o
r
r
ec
t p
r
ed
icti
o
n
s
[
2
5
]
.
T
h
e
r
esu
lts
in
T
ab
le
1
p
r
esen
t
a
co
m
p
ar
ativ
e
ev
alu
atio
n
o
f
th
r
ee
f
ea
tu
r
e
s
elec
tio
n
ap
p
r
o
ac
h
es:
th
e
W
r
ap
p
er
m
eth
o
d
,
th
e
E
m
b
e
d
d
ed
m
eth
o
d
,
an
d
th
e
Hy
b
r
id
m
eth
o
d
.
T
h
e
E
v
alu
atio
n
was
co
n
d
u
cted
to
d
eter
m
in
e
wh
ich
tec
h
n
iq
u
e
o
f
f
er
s
th
e
b
est
g
en
e
r
aliza
tio
n
p
er
f
o
r
m
an
ce
wh
e
n
ap
p
lied
to
im
b
alan
ce
d
,
h
ig
h
-
d
im
en
s
io
n
al
f
i
n
an
cial
d
ata.
I
n
th
e
c
o
n
tex
t
o
f
cr
ed
it
r
is
k
m
o
d
elin
g
,
s
elec
ted
f
ea
tu
r
es
m
u
s
t
s
u
p
p
o
r
t
b
o
th
h
ig
h
m
o
d
el
f
id
elity
d
u
r
in
g
tr
ain
in
g
an
d
s
u
s
tain
ed
p
r
e
d
ictiv
e
r
eliab
ilit
y
wh
en
ap
p
lied
t
o
p
r
e
v
io
u
s
l
y
u
n
s
ee
n
d
ata.
Alth
o
u
g
h
all
th
r
ee
m
eth
o
d
s
d
em
o
n
s
tr
ate
s
tr
o
n
g
p
e
r
f
o
r
m
a
n
ce
ac
r
o
s
s
th
e
e
v
alu
atio
n
m
etr
ics,
th
e
h
y
b
r
id
ap
p
r
o
ac
h
d
em
o
n
s
tr
ates
th
e
s
tr
o
n
g
est
g
en
er
aliza
tio
n
am
o
n
g
all
m
eth
o
d
s
,
with
a
m
in
im
al
p
er
f
o
r
m
an
ce
g
ap
b
etwe
en
t
r
ain
in
g
a
n
d
test
in
g
s
ets
an
d
t
h
e
f
ewe
s
t
m
is
clas
s
if
icatio
n
s
,
m
ak
in
g
it
th
e
m
o
s
t
d
ep
en
d
ab
le
ch
o
ice
f
o
r
d
e
p
lo
y
m
e
n
t in
p
r
ac
tical
s
ettin
g
s
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8
7
7
6
I
n
t J I
n
f
&
C
o
m
m
u
n
T
ec
h
n
o
l
,
Vo
l.
1
5
,
No
.
4
,
D
e
c
em
b
er
20
2
6
:
1
6
1
4
-
1
6
2
3
1620
T
h
e
W
r
ap
p
er
h
as
a
r
ec
all
o
f
1
.
0
0
0
0
f
o
r
b
o
th
th
e
tr
ain
an
d
test
s
e
t.
Desp
ite
th
is
,
th
e
s
m
al
l
test
-
tr
ain
p
r
ec
is
io
n
g
a
p
(
0
.
0
0
1
8
)
in
d
icat
es
s
o
m
e
s
lig
h
t
o
v
e
r
f
itti
n
g
.
T
h
e
W
r
ap
p
er
m
eth
o
d
ass
ess
es
t
h
e
ef
f
ec
t
o
f
f
ea
tu
r
e
s
u
b
s
ets
o
n
p
r
ed
ictio
n
p
e
r
f
o
r
m
an
ce
d
ir
ec
tly
.
Fo
r
th
is
r
ea
s
o
n
,
it
s
elec
ted
o
n
ly
o
n
e
f
e
atu
r
e
f
o
r
p
o
r
tf
o
li
o
class
if
icatio
n
.
T
ab
le
1
.
C
o
m
p
a
r
ativ
e
r
esu
lts
o
f
f
ea
tu
r
e
s
elec
tio
n
m
eth
o
d
s
Met
ri
c
W
ra
p
p
er
E
m
b
e
d
d
e
d
H
y
b
ri
d
N
o
n
e
T
ra
i
n
T
e
s
t
D
i
ffer
en
ce
T
ra
i
n
T
e
s
t
D
i
ffer
en
ce
T
ra
i
n
T
e
s
t
D
i
ffer
en
ce
T
ra
i
n
T
e
s
t
D
i
ffer
e
n
ce
A
cc
u
a
ry
0
,
9
8
0
3
0
,
9
7
9
3
0
,
0
0
1
0
0
,
9
8
4
0
0
,
9
8
3
0
0
,
0
0
1
0
0
,
9
8
3
9
0
,
9
8
4
0
0
,
0
0
0
1
0
,
5
0
0
4
0
,
5
0
0
0
0
,
0
0
0
4
Prec
i
s
i
o
n
0
,
9
6
2
1
0
,
9
6
0
3
0
,
0
0
1
8
0
,
9
7
0
0
0
,
9
6
8
9
0
,
0
0
1
1
0
,
9
7
0
0
0
,
9
6
9
6
0
,
0
0
0
4
0
,
5
0
0
2
0
,
5
0
0
0
0
,
0
0
0
2
Reca
l
l
1
,
0
0
0
0
1
,
0
0
0
0
-
0
,
9
9
8
9
0
,
9
9
8
0
0
,
0
0
0
9
0
,
9
9
8
6
0
,
9
9
9
3
0
,
0
0
0
7
1
,
0
0
0
0
1
,
0
0
0
0
-
F1
Sc
o
re
0
,
9
8
0
7
0
,
9
7
9
8
0
,
0
0
0
9
0
,
9
8
4
2
0
,
9
8
3
3
0
,
0
0
0
9
0
,
9
8
4
2
0
,
9
8
4
2
-
0
,
6
6
6
7
0
,
6
6
6
7
-
W
e
o
b
tain
a
h
ig
h
class
if
icatio
n
ac
cu
r
ac
y
b
u
t
at
th
e
ex
p
e
n
s
e
o
f
an
o
v
e
r
f
itted
m
o
d
el
with
a
test
-
tr
ain
d
if
f
er
en
ce
o
f
0
.
0
1
7
4
in
ac
c
u
r
ac
y
,
0
.
0
0
6
9
in
p
r
ec
is
io
n
,
an
d
0
.
0
0
4
1
in
F1
-
s
co
r
e.
T
h
is
r
esu
lt
is
lim
ited
b
y
th
e
in
tr
in
s
ic
ch
ar
ac
ter
is
tic
o
f
th
e
wr
ap
p
er
m
eth
o
d
th
at
te
n
d
s
to
o
v
er
f
it
o
n
th
e
tr
ai
n
in
g
d
ata
wh
en
p
er
f
o
r
m
in
g
f
ea
tu
r
e
s
elec
tio
n
,
lead
in
g
to
n
a
r
r
o
w
b
u
t
n
o
n
-
g
en
er
alis
ab
le
f
ea
tu
r
e
s
u
b
s
ets.
Mo
r
e
s
p
ec
if
ically
,
as th
e
ev
alu
atio
n
o
f
th
e
f
ea
tu
r
es
is
b
ased
o
n
th
e
o
u
t
-
of
-
s
am
p
le
p
r
ed
ictiv
e
p
er
f
o
r
m
an
ce
,
th
e
i
n
f
o
r
m
atio
n
p
o
ten
tially
r
e
tain
ed
b
y
s
elec
ted
f
ea
tu
r
es
is
s
o
lely
b
a
s
ed
o
n
th
e
d
ataset
at
h
an
d
,
t
h
u
s
cr
ea
tin
g
o
v
e
r
s
im
p
lifie
d
m
o
d
els
th
at
d
o
n
o
t
ac
co
u
n
t f
o
r
n
ew
o
r
d
i
f
f
er
en
t i
n
p
u
ts
.
T
h
e
E
m
b
ed
d
ed
m
eth
o
d
h
as
an
ac
cu
r
ac
y
o
f
0
.
9
8
3
0
a
n
d
s
m
all
test
-
tr
ain
d
if
f
er
en
ce
s
o
f
0
.
0
0
0
9
in
r
ec
all
an
d
F1
-
s
co
r
e,
s
u
g
g
esti
n
g
o
n
ly
a
s
lig
h
tly
o
v
er
f
itted
m
o
d
el.
T
h
e
f
ea
tu
r
e
s
elec
tio
n
is
p
er
f
o
r
m
ed
d
ir
ec
tly
i
n
th
e
tr
ain
in
g
p
r
o
ce
s
s
o
f
t
h
e
class
if
ier
u
s
in
g
r
e
g
u
lar
izatio
n
,
an
d
i
t
s
elec
ts
5
f
ea
tu
r
es
th
at
in
f
o
r
m
ab
o
u
t
th
e
ty
p
e
o
f
cr
ed
it,
esti
m
ated
in
te
r
est
in
c
o
m
e,
r
e
ce
n
cy
o
f
th
e
lo
an
as
s
ess
m
en
t,
th
e
ty
p
e
o
f
r
e
g
io
n
,
an
d
p
r
o
v
is
io
n
in
g
in
f
o
r
m
atio
n
.
T
h
is
r
esu
lts
in
a
less
n
ar
r
o
w
b
u
t
s
till
m
o
r
e
f
o
cu
s
ed
s
et
o
f
f
ea
tu
r
es
co
m
p
a
r
ed
to
th
e
b
aselin
e
m
o
d
el.
Ho
wev
e
r
,
b
ein
g
o
p
tim
ized
to
war
d
s
th
e
s
am
e
tr
ai
n
in
g
o
b
jectiv
es,
th
e
s
elec
tio
n
m
ay
s
till
en
co
d
e
non
-
p
er
ce
iv
ab
le
b
iases
p
r
ese
n
t
in
th
e
d
ata
at
h
a
n
d
,
th
u
s
s
lig
h
tly
af
f
ec
tin
g
th
e
m
o
d
el’
s
g
en
er
aliza
tio
n
p
er
f
o
r
m
an
ce
.
T
h
is
r
esu
lt
h
as
s
h
o
wn
a
lim
itatio
n
o
f
th
e
e
m
b
e
d
d
ed
f
ea
tu
r
e
s
elec
tio
n
m
et
h
o
d
as
im
p
lem
en
te
d
in
th
is
s
tu
d
y
:
d
esp
ite
im
p
r
o
v
in
g
ef
f
icien
c
y
b
y
lear
n
in
g
an
d
s
elec
tin
g
f
ea
tu
r
es
s
im
u
lta
n
eo
u
s
ly
,
its
s
tr
o
n
g
d
ep
en
d
e
n
ce
o
n
m
o
d
el
-
s
p
ec
if
i
c
o
b
jectiv
es
lim
its
its
ad
ap
ta
b
ilit
y
wh
en
f
ac
ed
with
d
i
f
f
er
en
t
o
r
u
n
s
ee
n
d
ata
d
is
tr
ib
u
tio
n
s
.
T
h
e
Hy
b
r
id
m
eth
o
d
h
as
th
e
b
est
p
er
f
o
r
m
an
ce
with
a
test
ac
cu
r
ac
y
o
f
0
.
9
8
4
0
an
d
a
tr
ain
in
g
ac
cu
r
ac
y
o
f
0
.
9
8
3
9
,
a
n
d
a
s
m
all
test
-
tr
a
in
d
if
f
e
r
en
ce
o
f
-
0
.
0
0
0
1
.
T
h
e
r
ec
all
is
v
er
y
h
ig
h
with
a
v
alu
e
o
f
0
.
9
9
9
3
an
d
th
e
F1
-
s
co
r
e
is
th
e
s
am
e
f
o
r
b
o
th
s
ets
at
0
.
9
8
4
2
.
T
h
is
m
et
h
o
d
also
s
h
o
ws
th
e
lo
west
n
u
m
b
er
o
f
m
is
class
if
icati
o
n
s
o
n
th
e
test
s
et.
T
h
ese
r
esu
lts
ca
n
b
e
ex
p
lain
e
d
b
y
th
e
two
-
s
tag
e
f
ea
tu
r
e
s
elec
tio
n
p
r
o
ce
s
s
th
at
co
m
b
in
es
a
f
ilter
in
g
ap
p
r
o
ac
h
in
t
h
e
f
o
r
m
o
f
L
ass
o
r
eg
u
lar
izatio
n
,
an
e
m
b
ed
d
e
d
m
eth
o
d
th
at
p
en
aliz
es
m
u
ltico
llin
ea
r
ity
an
d
n
o
is
e
in
th
e
d
ata,
an
d
Seq
u
en
tial
Fo
r
war
d
Selectio
n
,
th
e
wr
ap
p
er
m
et
h
o
d
th
at
r
an
k
s
f
e
atu
r
es
ac
co
r
d
in
g
to
th
eir
co
n
tr
i
b
u
tio
n
to
th
e
p
r
e
d
ictiv
e
o
b
jectiv
e.
T
h
e
s
elec
ted
f
ea
tu
r
es
ar
e
d
iv
er
s
e
a
n
d
i
n
f
o
r
m
ativ
e,
c
o
v
er
in
g
d
if
f
er
en
t
asp
ec
ts
o
f
th
e
f
i
n
an
c
ial
an
d
b
eh
a
v
io
r
al
c
h
ar
ac
ter
is
tics
o
f
th
e
lo
a
n
s
,
s
u
ch
as
in
ter
est
in
co
m
e,
cr
e
d
it
am
o
u
n
t,
i
n
s
tallm
en
t
v
alu
es,
d
u
r
atio
n
,
a
n
d
r
is
k
ass
ess
m
en
t.
As
a
r
esu
lt,
th
e
m
o
d
el
is
co
m
p
r
eh
en
s
iv
e
a
n
d
less
p
r
o
n
e
to
o
v
er
f
itti
n
g
o
r
u
n
d
er
f
itti
n
g
,
m
ea
n
in
g
th
at
it
is
b
o
t
h
ac
cu
r
ate
a
n
d
g
en
er
aliza
b
le.
T
h
e
r
esu
lts
o
b
tain
e
d
w
ith
th
e
Hy
b
r
i
d
ap
p
r
o
ac
h
in
d
icate
th
at
it
is
n
o
t
o
n
ly
we
ll
-
b
alan
ce
d
in
ter
m
s
o
f
m
o
d
el
co
m
p
lex
ity
an
d
r
elev
an
ce
,
b
u
t
th
at
it
is
also
a
b
le
to
c
o
m
b
in
e
d
if
f
er
en
t
s
tr
ateg
ies
o
f
s
elec
tio
n
in
a
s
y
n
e
r
g
is
tic
m
an
n
e
r
,
in
o
r
d
er
to
f
u
r
th
er
im
p
r
o
v
e
th
e
m
o
d
el’
s
g
en
er
aliza
tio
n
p
er
f
o
r
m
a
n
ce
,
esp
ec
ially
wh
en
wo
r
k
in
g
with
h
ig
h
-
d
im
en
s
io
n
al
an
d
im
b
alan
ce
d
f
in
a
n
cial
d
ata.
T
h
e
m
o
d
el
with
n
o
f
ea
t
u
r
e
s
elec
tio
n
h
as
t
h
e
wo
r
s
t
p
er
f
o
r
m
an
ce
o
v
er
all
o
n
all
ev
al
u
atio
n
m
et
r
ics.
T
h
e
test
-
tr
ain
d
if
f
er
e
n
ce
s
ar
e
t
h
e
lo
west
am
o
n
g
all
th
e
m
eth
o
d
s
b
u
t
th
e
ab
s
o
lu
te
v
alu
es
in
ac
cu
r
ac
y
,
p
r
ec
is
io
n
,
an
d
F1
-
s
co
r
e
a
r
e
also
t
h
e
lo
west,
an
d
th
e
r
e
is
a
v
e
r
y
h
ig
h
n
u
m
b
er
o
f
m
is
class
if
icatio
n
s
o
n
th
e
co
n
f
u
s
io
n
m
atr
ix
.
T
h
is
p
o
o
r
p
e
r
f
o
r
m
an
c
e
ca
n
b
e
ex
p
lain
ed
b
y
th
e
lac
k
o
f
f
ea
tu
r
e
s
elec
tio
n
th
at
lea
v
e
s
th
e
m
o
d
el
w
ith
all
th
e
av
ailab
le
f
ea
tu
r
es,
in
clu
d
i
n
g
ir
r
elev
an
t,
r
e
d
u
n
d
an
t,
o
r
n
o
is
y
o
n
es.
I
n
o
th
er
wo
r
d
s
,
th
is
r
esu
lts
in
a
lo
t
o
f
n
o
is
e
th
at
in
cr
ea
s
es
th
e
m
o
d
el
’
s
v
ar
ian
ce
a
n
d
d
ec
r
ea
s
es
th
e
clar
ity
o
f
th
e
p
r
ed
ictio
n
s
.
T
h
e
s
e
r
esu
lts
h
ig
h
lig
h
t
th
e
im
p
o
r
tan
ce
o
f
f
ea
tu
r
e
s
ele
ctio
n
wh
en
wo
r
k
in
g
with
co
m
p
lex
an
d
im
b
alan
ce
d
d
ata.
T
o
f
u
r
th
er
v
alid
ate
m
o
d
el
p
er
f
o
r
m
a
n
ce
,
5
-
f
o
l
d
cr
o
s
s
-
v
alid
atio
n
was
c
o
n
d
u
cted
o
n
t
h
e
L
o
g
is
tic
R
eg
r
ess
io
n
m
o
d
el
u
s
in
g
th
e
s
elec
ted
f
ea
tu
r
es.
T
h
e
r
esu
lts
s
h
o
w
r
elativ
ely
co
n
s
is
ten
t
ac
cu
r
ac
y
s
co
r
es
ac
r
o
s
s
f
o
ld
s
,
with
th
e
m
ea
n
p
er
f
o
r
m
an
ce
co
m
p
ar
a
b
le
to
th
e
Ho
ld
o
u
t
test
r
esu
lts
.
T
h
is
s
u
g
g
ests
th
at
th
e
m
o
d
el
p
er
f
o
r
m
an
ce
is
n
o
t
s
tr
o
n
g
ly
a
f
f
ec
ted
b
y
d
ata
p
ar
titi
o
n
in
g
,
i
n
d
icatin
g
r
ed
u
ce
d
r
is
k
o
f
o
p
t
im
is
tic
b
ias
in
th
e
ev
alu
atio
n
.
Alth
o
u
g
h
th
e
m
o
d
el
ac
h
iev
es
v
er
y
h
ig
h
r
ec
all
an
d
m
i
n
im
al
d
if
f
er
en
ce
s
b
etwe
en
tr
ain
i
n
g
a
n
d
test
in
g
p
er
f
o
r
m
an
ce
,
th
ese
r
esu
lts
ar
e
in
ter
p
r
eted
with
ca
u
tio
n
.
T
h
e
n
ea
r
-
p
e
r
f
ec
t
r
ec
all
m
ay
b
e
in
f
lu
en
ce
d
b
y
th
e
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J I
n
f
&
C
o
m
m
u
n
T
ec
h
n
o
l
I
SS
N:
2252
-
8
7
7
6
E
xp
lo
r
in
g
effec
tive
fea
tu
r
e
s
elec
tio
n
meth
o
d
s
in
lo
g
is
t
ic
r
eg
r
ess
io
n
fo
r
cred
it
…
(
B
r
ig
itta
P
erma
d
h
i
)
1621
d
ataset
ch
ar
ac
ter
is
tics
an
d
cl
ass
d
is
tr
ib
u
tio
n
.
T
h
er
ef
o
r
e,
m
o
d
el
p
er
f
o
r
m
an
ce
is
ev
alu
ate
d
n
o
t
o
n
ly
th
r
o
u
g
h
ac
cu
r
ac
y
m
etr
ics
b
u
t
also
th
r
o
u
g
h
m
is
class
if
icatio
n
an
aly
s
is
an
d
f
ea
tu
r
e
d
iv
er
s
ity
to
e
n
s
u
r
e
th
at
th
e
m
o
d
el
r
em
ain
s
g
en
er
aliza
b
le
a
n
d
r
eli
ab
le.
4.
CO
NCLU
SI
O
N
T
h
e
p
u
r
p
o
s
e
o
f
th
is
p
ap
er
is
to
im
p
r
o
v
e
th
e
cr
ed
it
e
v
alu
at
io
n
a
n
d
v
e
r
if
icatio
n
s
y
s
tem
u
s
ed
in
th
e
b
an
k
in
g
in
d
u
s
tr
y
t
o
d
eter
m
in
e
th
e
elig
ib
ilit
y
cr
iter
ia
f
o
r
c
r
e
d
it
ca
r
d
ap
p
r
o
v
al.
Fo
r
t
h
is
p
u
r
p
o
s
e,
it
is
ess
en
tial
to
s
elec
t
th
e
m
o
s
t
r
elev
an
t
f
ea
tu
r
es
to
en
s
u
r
e
r
eliab
le
an
d
s
u
cc
ess
f
u
l
d
ec
is
io
n
s
.
T
h
e
f
o
u
r
m
etr
ics
o
f
ac
cu
r
ac
y
,
p
r
ec
is
io
n
,
r
ec
all,
an
d
F1
-
s
co
r
e
ar
e
u
s
ed
as
t
h
e
p
r
im
ar
y
p
er
f
o
r
m
an
ce
in
d
icato
r
s
f
o
r
th
e
ML
m
o
d
els,
an
d
th
e
r
esu
lts
in
d
icate
th
at
th
e
Hy
b
r
i
d
m
o
d
el
was th
e
m
o
s
t g
en
er
ali
za
b
le
to
u
n
s
ee
n
d
ata.
T
h
e
r
esu
lts
f
o
r
th
e
m
o
d
el
th
at
d
id
n
o
t
p
er
f
o
r
m
f
ea
tu
r
e
s
elec
ti
o
n
wer
e
m
ain
ly
in
ten
d
e
d
to
h
ig
h
lig
h
t
th
e
r
is
k
s
o
f
u
s
in
g
all
av
ailab
le
v
ar
iab
les.
T
h
is
alg
o
r
ith
m
h
a
d
th
e
lo
west
p
er
f
o
r
m
an
ce
a
n
d
h
ig
h
est
n
u
m
b
er
s
o
f
f
alse p
o
s
itiv
es a
n
d
n
eg
ativ
es,
wh
ich
ca
n
b
e
d
ir
ec
tly
attr
ib
u
te
d
to
th
e
d
ata’
s
d
i
m
en
s
io
n
ality
,
im
b
alan
ce
,
an
d
t
h
e
p
r
esen
ce
o
f
n
o
is
e,
r
e
d
u
n
d
an
t,
an
d
u
n
r
elate
d
in
f
o
r
m
atio
n
.
T
h
is
clea
r
ly
in
d
icate
s
th
at
tar
g
et
ed
f
ea
tu
r
e
s
elec
tio
n
is
a
u
s
ef
u
l te
ch
n
iq
u
e
f
o
r
r
em
o
v
in
g
am
b
ig
u
ity
f
r
o
m
t
h
e
d
ata.
I
t
is
v
is
ib
le
th
at
t
h
e
th
r
e
e
m
et
h
o
d
s
h
a
v
e
s
ig
n
if
ica
n
t
ad
v
an
ta
g
es
an
d
d
is
ad
v
an
ta
g
e
s
:
th
e
h
ig
h
r
ec
all
o
f
th
e
W
r
ap
p
er
m
eth
o
d
d
ir
ec
tly
im
p
ac
ts
th
e
lo
w
p
er
f
o
r
m
an
ce
an
d
g
en
er
aliza
tio
n
o
f
th
e
r
esu
lts
.
Sin
ce
th
i
s
alg
o
r
ith
m
o
n
l
y
h
ad
o
n
e
f
ea
t
u
r
e
in
th
e
f
in
al
s
elec
tio
n
,
th
e
tr
ain
in
g
r
esu
lts
wer
e
s
u
b
ject
to
a
s
in
g
le
im
p
o
r
tan
t
in
f
lu
en
ce
an
d
h
ad
a
l
o
w
lev
e
l
o
f
b
ias.
T
h
e
e
m
b
ed
d
e
d
m
et
h
o
d
h
as
a
g
r
ea
ter
n
u
m
b
er
o
f
f
ea
tu
r
es,
an
d
wh
il
e
s
o
m
e
b
ias
is
also
v
is
ib
le
f
r
o
m
th
e
tr
ain
in
g
d
ata,
t
h
e
p
r
e
d
ictiv
e
an
d
r
elate
d
f
ea
tu
r
es
im
p
r
o
v
e
th
e
f
in
al
b
alan
ce
o
f
th
e
two
m
etr
ics.
B
o
th
o
f
th
ese
d
r
awb
ac
k
s
an
d
a
d
v
an
ta
g
e
s
ar
e
p
r
esen
t
in
th
e
hyb
r
i
d
m
e
th
o
d
,
as
it
u
s
es
th
e
E
m
b
ed
d
e
d
tech
n
iq
u
e
f
o
r
a
q
u
ick
in
itial
s
elec
tio
n
an
d
th
e
W
r
ap
p
er
m
eth
o
d
f
o
r
a
s
ec
o
n
d
ar
y
an
d
th
o
r
o
u
g
h
s
eq
u
en
tial
elim
in
atio
n
o
f
th
e
least
p
r
ed
ictiv
e
v
ar
iab
les.
T
h
u
s
,
th
e
o
b
tain
ed
m
o
d
el
h
as
a
s
u
f
f
icien
t
n
u
m
b
er
o
f
f
ea
tu
r
es
th
at
ar
e
d
iv
er
s
e
an
d
co
n
tex
tu
ally
u
n
iq
u
e,
m
a
k
in
g
th
is
alg
o
r
ith
m
th
e
m
o
s
t
s
u
cc
ess
f
u
l
an
d
th
e
m
o
s
t
d
ep
lo
y
a
b
le
in
r
ea
l
-
wo
r
ld
cr
e
d
it sco
r
in
g
.
I
n
ad
d
itio
n
t
o
its
s
tr
o
n
g
p
r
e
d
i
ctiv
e
p
er
f
o
r
m
an
ce
,
th
e
p
r
o
p
o
s
ed
Hy
b
r
id
m
o
d
el
d
em
o
n
s
tr
ate
s
p
r
ac
tical
ap
p
licab
il
ity
in
r
ea
l
-
wo
r
ld
b
an
k
in
g
e
n
v
ir
o
n
m
en
ts
.
B
y
le
v
er
ag
in
g
lo
g
is
tic
r
eg
r
ess
io
n
,
th
e
m
o
d
el
r
em
ain
s
h
ig
h
ly
i
n
ter
p
r
etab
le
an
d
alig
n
ed
with
r
eg
u
lat
o
r
y
e
x
p
ec
tatio
n
s
,
en
ab
lin
g
tr
an
s
p
ar
e
n
t
d
ec
is
io
n
-
m
ak
i
n
g
.
Fu
r
th
er
m
o
r
e
,
th
e
r
ed
u
ce
d
n
u
m
b
er
o
f
s
elec
ted
f
ea
tu
r
es c
o
n
tr
i
b
u
t
es to
co
m
p
u
tatio
n
al
ef
f
icie
n
cy
an
d
lo
wer
s
d
ata
co
llectio
n
co
s
ts
.
T
h
er
ef
o
r
e,
t
h
is
m
o
d
el
is
n
o
t
o
n
ly
ac
cu
r
ate
b
u
t
also
im
p
lem
en
tab
le,
m
ak
in
g
it
a
v
iab
le
d
ec
is
io
n
-
s
u
p
p
o
r
t to
o
l f
o
r
cr
e
d
i
t a
p
p
r
o
v
al
p
r
o
ce
s
s
es in
f
in
an
cial
in
s
titu
tio
n
s
.
T
h
is
s
tu
d
y
f
o
cu
s
es
p
r
im
ar
i
ly
o
n
th
e
p
r
ed
ictiv
e
m
o
d
elin
g
asp
ec
t
o
f
cr
e
d
it
s
co
r
in
g
an
d
d
o
es
n
o
t
ex
p
licitly
in
co
r
p
o
r
ate
r
eg
u
lato
r
y
f
r
a
m
ewo
r
k
s
s
u
ch
as
weig
h
t
o
f
ev
i
d
en
ce
(
W
OE
)
,
in
f
o
r
m
atio
n
v
alu
e
(
I
V)
,
o
r
f
air
n
ess
co
n
s
tr
ain
ts
.
Ho
we
v
er
,
th
e
u
s
e
o
f
L
o
g
is
tic
R
eg
r
ess
io
n
s
u
p
p
o
r
ts
in
ter
p
r
eta
b
ilit
y
an
d
alig
n
s
with
co
m
m
o
n
r
e
g
u
lato
r
y
ex
p
ec
tat
io
n
s
in
f
in
an
cial
in
s
titu
tio
n
s
.
Fu
tu
r
e
r
esear
ch
m
ay
ex
ten
d
th
is
wo
r
k
b
y
in
teg
r
atin
g
ex
p
lain
ab
ilit
y
tech
n
iq
u
es a
n
d
f
air
n
ess
co
n
s
id
er
at
io
n
s
.
ACK
NO
WL
E
DG
M
E
N
T
S
T
h
is
p
ap
er
also
co
n
tain
s
p
r
o
p
r
ietar
y
d
ata
,
in
clu
d
i
n
g
a
d
atas
et
an
d
m
ater
ials
u
n
d
er
a
n
o
n
-
d
is
clo
s
u
r
e
ag
r
ee
m
en
t
(
NDA)
.
Fo
r
th
at
r
ea
s
o
n
,
th
e
d
ataset
ca
n
n
o
t
b
e
d
is
tr
ib
u
ted
to
th
e
p
u
b
lic.
I
f
th
er
e
ar
e
an
y
d
ata
in
q
u
ir
ies
f
o
r
ac
ad
em
ic
p
u
r
p
o
s
es,
p
lease
s
en
d
th
e
in
q
u
i
r
y
to
t
h
e
au
th
o
r
v
ia
b
r
ig
itta.p
er
m
a
d
h
i@
b
in
u
s
.
ac
.
id
.
T
h
e
au
th
o
r
s
wo
u
ld
lik
e
to
a
c
k
n
o
wled
g
e
t
h
e
u
s
e
o
f
a
r
tific
ial
in
tellig
en
ce
to
o
ls
f
o
r
p
ar
ap
h
r
asin
g
an
d
lig
h
t
d
eb
u
g
g
in
g
ass
is
tan
ce
d
u
r
in
g
th
e
p
r
ep
a
r
atio
n
o
f
th
is
p
ap
er
.
All
f
in
al
in
te
r
p
r
etati
o
n
s
,
an
aly
s
es,
an
d
co
n
clu
s
io
n
s
ar
e
th
e
s
o
le
r
esp
o
n
s
ib
ilit
y
o
f
th
e
au
th
o
r
s
.
T
h
e
p
r
o
m
p
ts
u
s
ed
wer
e
:
i)
Please
p
ar
ap
h
r
ase
th
e
f
o
llo
win
g
p
ar
a
g
r
ap
h
to
m
ak
e
it
clea
r
an
d
u
n
d
er
s
tan
d
ab
le
an
d
m
ain
tain
th
e
s
am
e
m
ea
n
in
g
.
ii)
Please
h
elp
m
e
im
p
r
o
v
e
t
h
e
f
lu
e
n
cy
an
d
co
h
er
en
ce
o
f
th
is
p
a
r
t.
iii)
Please h
elp
im
p
r
o
v
e
co
d
e
clar
ity
an
d
ad
d
co
m
m
e
n
ts
.
F
UNDING
I
NF
O
R
M
A
T
I
O
N
T
h
e
au
th
o
r
s
s
tate
th
at
n
o
f
u
n
d
i
n
g
was
in
v
o
lv
e
d
.
AUTHO
R
CO
NT
RI
B
UT
I
O
NS ST
A
T
E
M
E
N
T
T
h
is
jo
u
r
n
al
u
s
es
th
e
C
o
n
tr
ib
u
to
r
R
o
les
T
ax
o
n
o
m
y
(
C
R
ed
iT)
to
r
ec
o
g
n
ize
in
d
iv
id
u
al
au
th
o
r
co
n
tr
ib
u
tio
n
s
,
r
ed
u
ce
au
th
o
r
s
h
ip
d
is
p
u
tes,
an
d
f
ac
ilit
ate
co
llab
o
r
atio
n
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8
7
7
6
I
n
t J I
n
f
&
C
o
m
m
u
n
T
ec
h
n
o
l
,
Vo
l.
1
5
,
No
.
4
,
D
e
c
em
b
er
20
2
6
:
1
6
1
4
-
1
6
2
3
1622
Na
m
e
o
f
Aut
ho
r
C
M
So
Va
Fo
I
R
D
O
E
Vi
Su
P
Fu
B
r
ig
itta Per
m
ad
h
i
✓
✓
✓
✓
✓
✓
✓
✓
✓
Dev
i Fitria
n
ah
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
C
:
C
o
n
c
e
p
t
u
a
l
i
z
a
t
i
o
n
M
:
M
e
t
h
o
d
o
l
o
g
y
So
:
So
f
t
w
a
r
e
Va
:
Va
l
i
d
a
t
i
o
n
Fo
:
Fo
r
mal
a
n
a
l
y
s
i
s
I
:
I
n
v
e
s
t
i
g
a
t
i
o
n
R
:
R
e
so
u
r
c
e
s
D
:
D
a
t
a
C
u
r
a
t
i
o
n
O
:
W
r
i
t
i
n
g
-
O
r
i
g
i
n
a
l
D
r
a
f
t
E
:
W
r
i
t
i
n
g
-
R
e
v
i
e
w
&
E
d
i
t
i
n
g
Vi
:
Vi
su
a
l
i
z
a
t
i
o
n
Su
:
Su
p
e
r
v
i
s
i
o
n
P
:
P
r
o
j
e
c
t
a
d
mi
n
i
st
r
a
t
i
o
n
Fu
:
Fu
n
d
i
n
g
a
c
q
u
i
si
t
i
o
n
CO
NF
L
I
C
T
O
F
I
N
T
E
R
E
S
T
ST
A
T
E
M
E
NT
T
h
e
au
th
o
r
s
s
tate
n
o
co
n
f
lict o
f
in
ter
est.
DATA AV
AI
L
AB
I
L
I
T
Y
T
h
e
d
ata
th
at
s
u
p
p
o
r
t
th
e
f
i
n
d
in
g
s
o
f
th
is
s
tu
d
y
ar
e
av
aila
b
le
o
n
r
eq
u
est
f
r
o
m
th
e
co
r
r
esp
o
n
d
in
g
au
th
o
r
,
[
i
n
itials
,
B
P].
T
h
e
d
at
a,
wh
ich
co
n
tain
in
f
o
r
m
atio
n
th
at
co
u
ld
co
m
p
r
o
m
is
e
th
e
p
r
iv
ac
y
o
f
r
es
ea
r
c
h
p
ar
ticip
an
ts
,
ar
e
n
o
t p
u
b
licly
a
v
ailab
le
d
u
e
to
ce
r
tain
r
estrictio
n
s
.
RE
F
E
R
E
NC
E
S
[
1
]
T.
K
o
b
a
y
a
s
h
i
e
t
a
l
.
,
“
D
e
v
e
l
o
p
m
e
n
t
o
f
a
sco
r
i
n
g
mo
d
e
l
b
a
se
d
o
n
o
b
j
e
c
t
i
v
e
f
a
c
t
o
r
s
t
o
p
r
e
d
i
c
t
g
a
n
g
r
e
n
o
u
s
/
p
e
r
f
o
r
a
t
e
d
a
p
p
e
n
d
i
c
i
t
i
s
,
”
BM
C
G
a
s
t
ro
e
n
t
e
r
o
l
o
g
y
,
v
o
l
.
2
3
,
n
o
.
1
,
2
0
2
3
,
d
o
i
:
1
0
.
1
1
8
6
/
s
1
2
8
7
6
-
023
-
0
2
7
6
7
-
7.
[
2
]
O
.
P
.
R
u
d
,
“
D
a
t
a
mi
n
i
n
g
c
o
o
k
b
o
o
k
:
mo
d
e
l
i
n
g
d
a
t
a
f
o
r
mar
k
e
t
i
n
g
,
r
i
s
k
,
a
n
d
c
u
st
o
mer
r
e
l
a
t
i
o
n
sh
i
p
ma
n
a
g
e
me
n
t
,
”
J
o
h
n
W
i
l
e
y
&
S
o
n
s
,
2
0
0
1
.
[
3
]
R
.
D
a
mi
a
z
a
a
n
d
D
.
F
i
t
r
i
a
n
a
h
,
“
P
r
e
d
i
c
t
i
o
n
a
n
a
l
y
si
s
o
f
K
a
r
t
u
Ja
k
a
r
t
a
P
i
n
t
a
r
(
K
JP
)
a
w
a
r
d
e
e
s
i
n
v
o
c
a
t
i
o
n
a
l
h
i
g
h
s
c
h
o
o
l
X
Y
Z
u
si
n
g
C
4
.
5
a
l
g
o
r
i
t
h
m,”
I
n
t
e
rn
a
t
i
o
n
a
l
J
o
u
rn
a
l
o
f
M
a
c
h
i
n
e
L
e
a
r
n
i
n
g
a
n
d
C
o
m
p
u
t
i
n
g
,
v
o
l
.
1
0
,
n
o
.
1
,
p
p
.
4
4
–
5
0
,
J
a
n
.
2
0
2
0
,
d
o
i
:
1
0
.
1
8
1
7
8
/
i
j
m
l
c
.
2
0
2
0
.
1
0
.
1
.
8
9
6
.
[
4
]
M
.
M
o
me
n
i
,
M
.
R
a
s
h
i
d
i
f
a
r
,
F
.
H
.
B
a
l
a
m
,
A
.
R
o
o
i
n
t
a
n
,
a
n
d
A
.
G
h
o
l
a
mi
n
e
j
a
d
,
“
A
c
o
m
p
r
e
h
e
n
s
i
v
e
a
n
a
l
y
s
i
s
o
f
g
e
n
e
e
x
p
r
e
ss
i
o
n
p
r
o
f
i
l
i
n
g
d
a
t
a
i
n
C
O
V
I
D
-
1
9
p
a
t
i
e
n
t
s
f
o
r
d
i
sc
o
v
e
r
y
o
f
sp
e
c
i
f
i
c
a
n
d
d
i
f
f
e
r
e
n
t
i
a
l
b
l
o
o
d
b
i
o
mar
k
e
r
si
g
n
a
t
u
r
e
s,”
S
c
i
e
n
t
i
f
i
c
Re
p
o
r
t
s
,
v
o
l
.
1
3
,
n
o
.
1
,
A
p
r
.
2
0
2
3
,
d
o
i
:
1
0
.
1
0
3
8
/
s
4
1
5
9
8
-
0
2
3
-
3
2
2
6
8
-
2.
[
5
]
J.
L
i
e
t
a
l
.
,
“
F
e
a
t
u
r
e
s
e
l
e
c
t
i
o
n
:
a
d
a
t
a
p
e
r
s
p
e
c
t
i
v
e
,
”
A
C
M
C
o
m
p
u
t
i
n
g
S
u
r
v
e
y
s
,
v
o
l
.
5
0
,
n
o
.
6
,
p
p
.
1
–
4
5
,
D
e
c
.
2
0
1
7
,
d
o
i
:
1
0
.
1
1
4
5
/
3
1
3
6
6
2
5
.
[
6
]
R
.
S
a
l
ma
n
,
A
.
A
l
z
a
a
t
r
e
h
,
a
n
d
M
.
T
.
A
l
B
a
t
a
i
n
e
h
,
“
F
e
a
t
u
r
e
s
e
l
e
c
t
i
o
n
o
f
t
h
e
r
e
s
p
i
r
a
t
o
r
y
m
i
c
r
o
b
i
o
t
a
a
ss
o
c
i
a
t
e
d
w
i
t
h
a
st
h
m
a
,
”
J
o
u
rn
a
l
o
f
Bi
g
D
a
t
a
,
v
o
l
.
1
0
,
n
o
.
1
,
2
0
2
3
,
d
o
i
:
1
0
.
1
1
8
6
/
s
4
0
5
3
7
-
023
-
0
0
7
6
7
-
8.
[
7
]
Y.
-
W
.
C
h
e
n
a
n
d
C
.
-
J.
L
i
n
,
“
C
o
m
b
i
n
i
n
g
S
V
M
s
w
i
t
h
v
a
r
i
o
u
s
f
e
a
t
u
r
e
se
l
e
c
t
i
o
n
s
t
r
a
t
e
g
i
e
s
,
”
i
n
F
e
a
t
u
re
E
x
t
r
a
c
t
i
o
n
,
S
p
r
i
n
g
e
r
B
e
r
l
i
n
H
e
i
d
e
l
b
e
r
g
,
p
p
.
3
1
5
–
3
2
4
.
[
8
]
V
.
K
u
mar,
“
F
e
a
t
u
r
e
se
l
e
c
t
i
o
n
:
a
l
i
t
e
r
a
t
u
r
e
r
e
v
i
e
w
,
”
T
h
e
S
m
a
rt
C
o
m
p
u
t
i
n
g
Re
v
i
e
w
,
v
o
l
.
4
,
n
o
.
3
,
2
0
1
4
,
d
o
i
:
1
0
.
6
0
2
9
/
s
mart
c
r
.
2
0
1
4
.
0
3
.
0
0
7
.
[
9
]
B
.
V
e
n
k
a
t
e
s
h
a
n
d
J.
A
n
u
r
a
d
h
a
,
“
A
r
e
v
i
e
w
o
f
f
e
a
t
u
r
e
s
e
l
e
c
t
i
o
n
a
n
d
i
t
s
met
h
o
d
s,
”
C
y
b
e
r
n
e
t
i
c
s
a
n
d
I
n
f
o
rm
a
t
i
o
n
T
e
c
h
n
o
l
o
g
i
e
s
,
v
o
l
.
1
9
,
n
o
.
1
,
p
p
.
3
–
2
6
,
M
a
r
.
2
0
1
9
,
d
o
i
:
1
0
.
2
4
7
8
/
c
a
i
t
-
2
0
1
9
-
0
0
0
1
.
[
1
0
]
S
.
A
.
A
l
a
sa
d
i
a
n
d
W
.
S
.
B
h
a
y
a
,
“
R
e
v
i
e
w
o
f
d
a
t
a
p
r
e
p
r
o
c
e
s
si
n
g
t
e
c
h
n
i
q
u
e
s
i
n
d
a
t
a
mi
n
i
n
g
,
”
J
o
u
r
n
a
l
o
f
En
g
i
n
e
e
r
i
n
g
a
n
d
A
p
p
l
i
e
d
S
c
i
e
n
c
e
s
,
v
o
l
.
1
2
,
n
o
.
1
6
,
p
p
.
4
1
0
2
–
4
1
0
7
,
2
0
1
7
.
[
1
1
]
D
.
A
.
V
.
d
e
P
a
u
l
a
,
R
.
A
r
t
e
s
,
F
.
A
y
r
e
s
,
a
n
d
A
.
M
.
A
.
F
.
M
i
n
a
r
d
i
,
“
Es
t
i
m
a
t
i
n
g
c
r
e
d
i
t
a
n
d
p
r
o
f
i
t
s
c
o
r
i
n
g
o
f
a
B
r
a
z
i
l
i
a
n
c
r
e
d
i
t
u
n
i
o
n
w
i
t
h
l
o
g
i
st
i
c
r
e
g
r
e
ssi
o
n
a
n
d
ma
c
h
i
n
e
-
l
e
a
r
n
i
n
g
t
e
c
h
n
i
q
u
e
s
,
”
RAU
S
P
M
a
n
a
g
e
m
e
n
t
J
o
u
rn
a
l
,
v
o
l
.
5
4
,
n
o
.
3
,
p
p
.
3
2
1
–
3
3
6
,
2
0
1
9
,
d
o
i
:
1
0
.
1
1
0
8
/
r
a
u
s
p
-
03
-
2
0
1
8
-
0
0
0
3
.
[
1
2
]
R
.
G
a
i
a
n
d
H
.
Z
h
a
n
g
,
“
P
r
e
d
i
c
t
i
o
n
mo
d
e
l
o
f
a
g
r
i
c
u
l
t
u
r
a
l
w
a
t
e
r
q
u
a
l
i
t
y
b
a
se
d
o
n
o
p
t
i
mi
z
e
d
l
o
g
i
s
t
i
c
r
e
g
r
e
s
si
o
n
a
l
g
o
r
i
t
h
m,
”
EU
RAS
I
P
J
o
u
rn
a
l
o
n
Ad
v
a
n
c
e
s
i
n
S
i
g
n
a
l
Pr
o
c
e
ss
i
n
g
,
v
o
l
.
2
0
2
3
,
n
o
.
1
,
F
e
b
.
2
0
2
3
,
d
o
i
:
1
0
.
1
1
8
6
/
s1
3
6
3
4
-
0
2
3
-
0
0
9
7
3
-
9.
[
1
3
]
D
.
F
i
t
r
i
a
n
a
h
,
“
F
e
a
t
u
r
e
e
x
p
l
o
r
a
t
i
o
n
f
o
r
p
r
e
d
i
c
t
i
o
n
o
f
p
o
t
e
n
t
i
a
l
t
u
n
a
f
i
s
h
i
n
g
z
o
n
e
s,”
I
n
t
e
rn
a
t
i
o
n
a
l
J
o
u
r
n
a
l
o
f
I
n
f
o
rm
a
t
i
o
n
a
n
d
El
e
c
t
r
o
n
i
c
s
E
n
g
i
n
e
e
r
i
n
g
,
2
0
1
5
,
d
o
i
:
1
0
.
7
7
6
3
/
i
j
i
e
e
.
2
0
1
5
.
v
5
.
5
4
3
.
[
1
4
]
M
.
J.
Er
sh
a
d
i
a
n
d
D
.
O
m
i
d
z
a
d
e
h
,
“
C
u
st
o
mer
v
a
l
i
d
a
t
i
o
n
u
s
i
n
g
h
y
b
r
i
d
l
o
g
i
s
t
i
c
r
e
g
r
e
ss
i
o
n
a
n
d
c
r
e
d
i
t
sc
o
r
i
n
g
m
o
d
e
l
:
a
c
a
se
st
u
d
y
,
”
C
a
l
i
t
a
t
e
a
,
v
o
l
.
1
9
,
n
o
.
1
6
7
,
p
p
.
5
9
–
6
2
,
2
0
1
8
.
[
1
5
]
Z.
Y
e
,
H
.
Z
h
a
n
g
,
J.
L
i
a
n
g
,
S
.
Y
i
,
a
n
d
X
.
Z
h
a
n
,
“
S
i
g
n
i
f
i
c
a
n
c
e
o
f
l
o
g
i
s
t
i
c
r
e
g
r
e
ssi
o
n
sc
o
r
i
n
g
m
o
d
e
l
b
a
se
d
o
n
n
a
t
u
r
a
l
k
i
l
l
e
r
c
e
l
l
-
me
d
i
a
t
e
d
c
y
t
o
t
o
x
i
c
p
a
t
h
w
a
y
i
n
t
h
e
d
i
a
g
n
o
si
s
o
f
c
o
l
o
n
c
a
n
c
e
r
,
”
Fr
o
n
t
i
e
rs
i
n
I
m
m
u
n
o
l
o
g
y
,
v
o
l
.
1
4
,
J
a
n
.
2
0
2
3
,
d
o
i
:
1
0
.
3
3
8
9
/
f
i
mm
u
.
2
0
2
3
.
1
1
1
7
9
0
8
.
[
1
6
]
A
.
R
a
h
ma
n
,
“
S
t
a
t
i
st
i
c
s
-
b
a
se
d
d
a
t
a
p
r
e
p
r
o
c
e
s
si
n
g
me
t
h
o
d
s
a
n
d
ma
c
h
i
n
e
l
e
a
r
n
i
n
g
a
l
g
o
r
i
t
h
ms
f
o
r
b
i
g
d
a
t
a
a
n
a
l
y
si
s
,
”
I
n
t
e
r
n
a
t
i
o
n
a
l
J
o
u
r
n
a
l
o
f
Ar
t
i
f
i
c
i
a
l
I
n
t
e
l
l
i
g
e
n
c
e
,
v
o
l
.
1
7
,
n
o
.
2
,
p
p
.
4
4
–
6
5
,
2
0
1
9
.
[
1
7
]
M
.
B
e
n
n
a
sar,
Y
.
H
i
c
k
s
,
a
n
d
R
.
S
e
t
c
h
i
,
“
F
e
a
t
u
r
e
sel
e
c
t
i
o
n
u
si
n
g
j
o
i
n
t
m
u
t
u
a
l
i
n
f
o
r
m
a
t
i
o
n
ma
x
i
mi
s
a
t
i
o
n
,
”
Ex
p
e
rt
S
y
st
e
m
s
w
i
t
h
Ap
p
l
i
c
a
t
i
o
n
s
,
v
o
l
.
4
2
,
n
o
.
2
2
,
p
p
.
8
5
2
0
–
8
5
3
2
,
D
e
c
.
2
0
1
5
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
e
sw
a
.
2
0
1
5
.
0
7
.
0
0
7
.
[
1
8
]
J.
S
u
b
r
a
m
a
n
i
a
n
a
n
d
R
.
S
i
mo
n
,
“
O
v
e
r
f
i
t
t
i
n
g
i
n
p
r
e
d
i
c
t
i
o
n
m
o
d
e
l
s
–
I
s
i
t
a
p
r
o
b
l
e
m
o
n
l
y
i
n
h
i
g
h
d
i
me
n
si
o
n
s?
,
”
C
o
n
t
e
m
p
o
r
a
ry
C
l
i
n
i
c
a
l
T
ri
a
l
s
,
v
o
l
.
3
6
,
n
o
.
2
,
p
p
.
6
3
6
–
6
4
1
,
N
o
v
.
2
0
1
3
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
c
c
t
.
2
0
1
3
.
0
6
.
0
1
1
.
[
1
9
]
O
.
M
e
h
r
p
o
u
r
e
t
a
l
.
,
“
C
o
m
p
a
r
i
s
o
n
o
f
d
e
c
i
s
i
o
n
t
r
e
e
w
i
t
h
c
o
m
mo
n
mac
h
i
n
e
l
e
a
r
n
i
n
g
mo
d
e
l
s
f
o
r
p
r
e
d
i
c
t
i
o
n
o
f
b
i
g
u
a
n
i
d
e
a
n
d
su
l
f
o
n
y
l
u
r
e
a
p
o
i
s
o
n
i
n
g
i
n
t
h
e
U
n
i
t
e
d
S
t
a
t
e
s
:
a
n
a
n
a
l
y
s
i
s
o
f
t
h
e
N
a
t
i
o
n
a
l
P
o
i
s
o
n
D
a
t
a
S
y
s
t
e
m,
”
BM
C
M
e
d
i
c
a
l
I
n
f
o
rm
a
t
i
c
s
a
n
d
D
e
c
i
s
i
o
n
Ma
k
i
n
g
,
v
o
l
.
2
3
,
n
o
.
1
,
A
p
r
.
2
0
2
3
,
d
o
i
:
1
0
.
1
1
8
6
/
s1
2
9
1
1
-
0
2
2
-
0
2
0
9
5
-
y.
[
2
0
]
K
.
H
.
H
u
i
,
C
.
S
.
O
o
i
,
M
.
H
.
L
i
m,
M
.
S
.
Le
o
n
g
,
a
n
d
S
.
M
.
A
l
-
O
b
a
i
d
i
,
“
A
n
i
m
p
r
o
v
e
d
w
r
a
p
p
e
r
-
b
a
s
e
d
f
e
a
t
u
r
e
s
e
l
e
c
t
i
o
n
m
e
t
h
o
d
f
o
r
mac
h
i
n
e
r
y
f
a
u
l
t
d
i
a
g
n
o
si
s
,
”
PLO
S
O
N
E
,
v
o
l
.
1
2
,
n
o
.
1
2
,
p
.
e
0
1
8
9
1
4
3
,
D
e
c
.
2
0
1
7
,
d
o
i
:
1
0
.
1
3
7
1
/
j
o
u
r
n
a
l
.
p
o
n
e
.
0
1
8
9
1
4
3
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J I
n
f
&
C
o
m
m
u
n
T
ec
h
n
o
l
I
SS
N:
2252
-
8
7
7
6
E
xp
lo
r
in
g
effec
tive
fea
tu
r
e
s
elec
tio
n
meth
o
d
s
in
lo
g
is
t
ic
r
eg
r
ess
io
n
fo
r
cred
it
…
(
B
r
ig
itta
P
erma
d
h
i
)
1623
[
2
1
]
Y
.
B
.
W
a
h
,
N
.
I
b
r
a
h
i
m,
H
.
A
.
H
a
mi
d
,
S
.
A
b
d
u
l
-
R
a
h
ma
n
,
a
n
d
S
.
F
o
n
g
,
“
F
e
a
t
u
r
e
se
l
e
c
t
i
o
n
me
t
h
o
d
s:
c
a
se
o
f
f
i
l
t
e
r
a
n
d
w
r
a
p
p
e
r
a
p
p
r
o
a
c
h
e
s f
o
r
m
a
x
i
mi
s
i
n
g
c
l
a
ssi
f
i
c
a
t
i
o
n
a
c
c
u
r
a
c
y
,
”
Pe
r
t
a
n
i
k
a
J
o
u
rn
a
l
o
f
S
c
i
e
n
c
e
a
n
d
T
e
c
h
n
o
l
o
g
y
,
v
o
l
.
2
6
,
n
o
.
1
,
p
.
3
2
9
,
2
0
1
8
.
[
2
2
]
U
.
M
.
K
h
a
i
r
e
a
n
d
R
.
D
h
a
n
a
l
a
k
sh
mi
,
“
S
t
a
b
i
l
i
t
y
o
f
f
e
a
t
u
r
e
se
l
e
c
t
i
o
n
a
l
g
o
r
i
t
h
m:
A
r
e
v
i
e
w
,
”
J
o
u
r
n
a
l
o
f
K
i
n
g
S
a
u
d
U
n
i
v
e
rs
i
t
y
-
C
o
m
p
u
t
e
r
a
n
d
I
n
f
o
rm
a
t
i
o
n
S
c
i
e
n
c
e
s
,
v
o
l
.
3
4
,
n
o
.
4
,
p
p
.
1
0
6
0
–
1
0
7
3
,
A
p
r
.
2
0
2
2
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
j
k
s
u
c
i
.
2
0
1
9
.
0
6
.
0
1
2
.
[
2
3
]
X.
-
Y
.
Li
u
,
Y
.
Li
a
n
g
,
S
.
W
a
n
g
,
Z
.
-
Y
.
Y
a
n
g
,
a
n
d
H
.
-
S
.
Y
e
,
“
A
h
y
b
r
i
d
g
e
n
e
t
i
c
a
l
g
o
r
i
t
h
m
w
i
t
h
w
r
a
p
p
e
r
-
e
m
b
e
d
d
e
d
a
p
p
r
o
a
c
h
e
s
f
o
r
f
e
a
t
u
r
e
s
e
l
e
c
t
i
o
n
,
”
I
E
EE
Ac
c
e
ss
,
v
o
l
.
6
,
p
p
.
2
2
8
6
3
–
2
2
8
7
4
,
2
0
1
8
,
d
o
i
:
1
0
.
1
1
0
9
/
a
c
c
e
s
s.
2
0
1
8
.
2
8
1
8
6
8
2
.
[
2
4
]
J.
L
i
u
e
t
a
l
.
,
“
D
e
v
e
l
o
p
m
e
n
t
a
n
d
v
a
l
i
d
a
t
i
o
n
o
f
p
r
e
d
i
c
t
i
v
e
m
o
d
e
l
b
a
se
d
o
n
d
e
e
p
l
e
a
r
n
i
n
g
m
e
t
h
o
d
f
o
r
c
l
a
ss
i
f
i
c
a
t
i
o
n
o
f
d
y
sl
i
p
i
d
e
m
i
a
i
n
C
h
i
n
e
se
me
d
i
c
i
n
e
,
”
H
e
a
l
t
h
I
n
f
o
rm
a
t
i
o
n
S
c
i
e
n
c
e
a
n
d
S
y
s
t
e
m
s
,
v
o
l
.
1
1
,
n
o
.
1
,
A
p
r
.
2
0
2
3
,
d
o
i
:
1
0
.
1
0
0
7
/
s1
3
7
5
5
-
0
2
3
-
0
0
2
1
5
-
0.
[
2
5
]
N
.
K
u
s
h
mer
i
c
k
,
“
W
r
a
p
p
e
r
i
n
d
u
c
t
i
o
n
:
Ef
f
i
c
i
e
n
c
y
a
n
d
e
x
p
r
e
ssi
v
e
n
e
ss,”
A
rt
i
f
i
c
i
a
l
I
n
t
e
l
l
i
g
e
n
c
e
,
v
o
l
.
1
1
8
,
n
o
.
1
–
2
,
p
p
.
15
–
6
8
,
A
p
r
.
2
0
0
0
,
d
o
i
:
1
0
.
1
0
1
6
/
s0
0
0
4
-
3
7
0
2
(
9
9
)
0
0
1
0
0
-
9.
[
2
6
]
R
.
K
o
h
a
v
i
a
n
d
G
.
H
.
J
o
h
n
,
“
W
r
a
p
p
e
r
s
f
o
r
f
e
a
t
u
r
e
su
b
s
e
t
se
l
e
c
t
i
o
n
,
”
A
rt
i
f
i
c
i
a
l
I
n
t
e
l
l
i
g
e
n
c
e
,
v
o
l
.
9
7
,
n
o
.
1
–
2
,
p
p
.
2
7
3
–
3
2
4
,
D
e
c
.
1
9
9
7
,
d
o
i
:
1
0
.
1
0
1
6
/
s
0
0
0
4
-
3
7
0
2
(
9
7
)
0
0
0
4
3
-
x.
B
I
O
G
RAP
H
I
E
S O
F
AUTH
O
RS
Br
ig
itta
Per
m
a
d
h
i
h
a
s
a
Ba
c
h
e
l
o
r
o
f
C
o
m
p
u
ter
S
c
ien
c
e
(S
.
K
o
m
.
)
,
wit
h
t
h
e
c
las
s
o
f
2
0
2
4
a
n
d
a
M
a
ste
r
o
f
Co
m
p
u
ter
S
c
ien
c
e
(M
.
Ko
m
.
)
fr
o
m
Bin
a
Nu
sa
n
tara
Un
iv
e
rsit
y
(Bin
u
s
),
In
d
o
n
e
sia
.
Bri
g
it
ta
P
e
rm
a
d
h
i'
s
field
o
f
st
u
d
y
is
c
o
m
p
u
ter
sc
ien
c
e
with
a
n
e
m
p
h
a
sis o
n
d
a
ta
sc
ien
c
e
a
n
d
n
a
tu
ra
l
la
n
g
u
a
g
e
p
ro
c
e
ss
in
g
(NLP
).
He
r
re
se
a
rc
h
in
tere
sts
in
c
lu
d
e
d
a
ta
sc
ien
c
e
,
n
a
tu
ra
l
lan
g
u
a
g
e
p
r
o
c
e
ss
in
g
,
a
n
d
a
p
p
l
ica
ti
o
n
s
wi
th
i
n
t
h
e
field
o
f
c
o
m
p
u
ter
sc
ien
c
e
.
S
h
e
h
a
s
b
e
e
n
i
n
v
o
lv
e
d
in
re
se
a
rc
h
p
r
o
jec
ts
o
r
a
c
a
d
e
m
ic
wo
rk
re
late
d
to
h
y
b
r
id
a
p
p
ro
a
c
h
e
s
fo
r
tex
t
su
m
m
a
riza
ti
o
n
u
si
n
g
BERT
a
n
d
T5
m
o
d
e
ls,
u
ti
li
z
in
g
th
e
t
ra
n
sfo
rm
e
rs
li
b
ra
ry
in
P
y
th
o
n
.
S
h
e
c
a
n
b
e
c
o
n
tac
ted
a
t
e
m
a
il
:
b
ri
g
it
ta.p
e
rm
a
d
h
i@b
i
n
u
s.a
c
.
id
.
De
v
i
Fi
tr
ia
n
a
h
is
a
lec
tu
re
r
a
n
d
re
se
a
rc
h
e
r
in
M
a
ste
r
o
f
C
o
m
p
u
ter
S
c
ien
c
e
De
p
a
rtme
n
t
in
Bin
a
N
u
sa
n
tara
Un
iv
e
rsit
y
(Bin
u
s
),
In
d
o
n
e
sia
.
S
h
e
re
c
e
iv
e
d
Ba
c
h
e
lo
r’s
d
e
g
re
e
in
Co
m
p
u
ter
S
c
ien
c
e
m
a
jo
r
i
n
C
o
m
p
u
ter
S
c
ien
c
e
fro
m
Bin
a
Nu
s
a
n
tara
Un
iv
e
rsity
in
2
0
0
0
,
M
a
ste
r’s
d
e
g
re
e
in
I
n
fo
rm
a
ti
o
n
Tec
h
n
o
l
o
g
y
fro
m
U
n
iv
e
rsita
s
In
d
o
n
e
sia
in
2
0
0
8
,
a
n
d
Do
c
to
r
o
f
P
h
il
o
so
p
h
y
(P
h
.
D.)
i
n
Co
m
p
u
ter
S
c
ien
c
e
fro
m
Un
iv
e
rsi
tas
In
d
o
n
e
sia
in
2
0
1
5
.
S
h
e
re
c
e
iv
e
d
re
se
a
rc
h
p
lac
e
m
e
n
t
(S
a
n
d
wic
h
P
ro
g
ra
m
)
in
Lab
o
ra
to
ry
f
o
r
P
a
tt
e
rn
Re
c
o
g
n
it
i
o
n
a
n
d
Im
a
g
e
P
ro
c
e
ss
in
g
a
n
d
G
IS
(P
RI
P
G
IS
Lab
),
De
p
a
rtme
n
t
o
f
Co
m
p
u
t
e
r
S
c
ien
c
e
,
M
ich
i
g
a
n
S
tate
Un
iv
e
rsity
,
Eas
t
Lan
sin
g
,
M
ic
h
i
g
a
n
,
Un
i
ted
S
tate
s,
in
2
0
1
4
.
S
h
e
is
c
u
rre
n
tl
y
a
c
ti
v
e
in
th
e
lec
tu
rin
g
a
n
d
re
se
a
rc
h
e
s
in
t
h
e
field
o
f
d
a
ta
m
in
i
n
g
,
b
ig
d
a
t
a
a
n
a
ly
ti
c
s,
a
n
d
a
rti
ficia
l
in
telli
g
e
n
c
e
.
Th
e
c
o
u
rse
s
sh
e
tea
c
h
e
s
in
c
lu
d
e
d
a
ta
m
in
in
g
,
b
ig
d
a
ta
a
n
a
ly
ti
c
s,
a
n
d
to
p
ics
in
a
rti
ficia
l
in
tell
ig
e
n
c
e
.
He
r
re
se
a
rc
h
in
tere
sts
in
c
lu
d
e
d
a
ta
m
in
i
n
g
,
m
a
c
h
in
e
lea
rn
i
n
g
,
a
rti
ficia
l
in
telli
g
e
n
c
e
,
a
n
d
a
p
p
l
ied
re
m
o
te
se
n
sin
g
.
S
h
e
c
a
n
b
e
c
o
n
tac
ted
a
t
e
m
a
il
:
d
e
v
i.
fit
r
ian
a
h
@b
in
u
s.a
c
.
id
.
Evaluation Warning : The document was created with Spire.PDF for Python.