I
nd
o
ne
s
ia
n J
o
urna
l o
f
E
lect
rica
l En
g
ineering
a
nd
Co
m
p
u
t
er
Science
Vo
l.
17
,
No
.
1
,
J
an
u
ar
y
20
20
,
p
p
.
5
4
3
~5
5
0
I
SS
N:
2
5
0
2
-
4
7
5
2
,
DOI
: 1
0
.
1
1
5
9
1
/i
j
ee
cs.v
1
7
.i
1
.
p
p
5
4
3
-
5
5
0
543
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
/
/
ijeec
s
.
ia
es
co
r
e.
co
m
Propo
sed a
lg
o
rith
m
for
R
eg
r
ess
io
n
-
ba
sed predic
tion
w
ith
bul
k
nois
e
Cha
nin
t
o
rn
J
it
t
a
w
iriy
a
nu
k
o
o
n
G
ra
d
u
a
te S
c
h
o
o
l
o
f
A
d
v
a
n
c
e
d
T
e
c
h
n
o
l
o
g
y
M
a
n
a
g
e
m
e
n
t,
A
ss
u
m
p
ti
o
n
U
n
iv
e
rsity
,
T
h
a
il
a
n
d
Art
icle
I
nfo
AB
ST
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
A
p
r
22
,
2
0
1
9
R
ev
i
s
ed
J
u
n
2
3
,
2
0
1
9
A
cc
ep
ted
J
u
l
1
,
2
0
1
9
T
h
e
n
o
ise
h
a
s
in
c
it
e
d
a
n
o
rig
in
a
l
d
a
ta
d
u
e
to
a
n
e
tw
o
rk
w
it
h
a
n
in
fe
rio
r
S
NR.
In
c
a
se
o
f
th
e
b
u
lk
n
o
ise
,
th
e
in
sig
h
tf
u
l
c
o
n
ten
t
w
it
h
in
th
e
d
a
ta
i
s
su
b
sta
n
ti
a
ll
y
sq
u
e
e
z
e
d
.
A
c
o
st
-
e
ff
e
c
ti
v
e
m
e
th
o
d
w
il
l
c
h
a
ll
e
n
g
e
to
q
u
a
ra
n
ti
n
e
th
e
in
sig
h
ts,
so
th
a
t
in
f
o
rm
a
ti
o
n
c
a
n
b
e
u
ti
li
z
e
d
m
o
re
re
so
u
rc
e
f
u
ll
y
.
T
o
a
c
h
iev
e
th
is
a
i
m
,
it
is
e
ss
e
n
ti
a
l
to
iro
n
th
e
b
u
lk
n
o
ise
c
o
n
ten
t
o
u
t
,
a
n
d
t
h
e
n
c
a
lcu
late
th
e
a
n
a
l
y
t
ic
s
o
f
th
e
c
le
a
n
d
a
ta.
A
s
n
o
ise
is
b
u
lk
so
so
m
e
sta
ti
stica
l
m
e
th
o
d
o
lo
g
ies
su
c
h
a
s
a
v
e
ra
g
in
g
o
r
ra
n
d
o
m
izin
g
a
re
e
m
p
lo
y
e
d
.
A
p
re
d
ictio
n
u
sin
g
th
e
re
g
re
ss
io
n
-
b
a
se
d
m
o
d
e
l
w
it
h
b
u
lk
n
o
ise
f
o
r
t
h
e
e
x
p
e
rime
n
t
in
p
ra
c
ti
c
e
is
in
tro
d
u
c
e
d
.
T
h
e
d
e
c
o
m
p
o
siti
o
n
a
p
p
ro
a
c
h
t
o
se
p
a
ra
te t
h
e
in
sig
h
ts
is
e
x
p
lo
it
e
d
.
T
h
e
p
r
o
p
o
se
d
a
lg
o
ri
th
m
a
c
h
iev
e
s
a
(lo
c
a
l)
so
lu
ti
o
n
a
t
e
a
c
h
c
o
m
p
u
ti
n
g
ste
p
a
n
d
se
lec
ts
th
e
b
e
st
so
lu
ti
o
n
in
v
iew
o
f
g
lo
b
a
l
i
m
p
a
c
ts.
T
h
e
c
o
rre
latio
n
c
o
e
f
f
icie
n
t,
a
v
e
r
a
g
e
e
rro
r,
a
b
so
lu
te
e
rro
r
a
n
d
m
e
a
n
sq
u
a
re
d
e
rro
r
a
re
u
se
d
to
c
o
n
stit
u
te
th
e
p
re
d
ict
io
n
.
Re
su
lt
s
f
ro
m
M
OA
si
m
u
latio
n
w
il
l
b
e
c
o
m
p
a
re
d
to
a
c
tu
a
l
d
a
ta
in
t
h
e
su
c
c
e
e
d
in
g
ti
m
e
.
T
h
e
p
re
d
ictio
n
w
it
h
b
u
lk
n
o
ise
u
si
n
g
th
e
p
ro
p
o
se
d
a
lg
o
rit
h
m
o
u
tp
e
rf
o
r
m
s o
th
e
r
im
p
u
tatio
n
m
e
th
o
d
s.
K
ey
w
o
r
d
s
:
B
u
lk
n
o
is
e
Me
an
ab
s
o
lu
te
er
r
o
r
Me
an
s
q
u
ar
e
er
r
o
r
No
is
y
a
n
d
m
i
s
s
i
n
g
d
ata
R
eg
r
es
s
io
n
-
b
ased
p
r
ed
ictio
n
Sig
n
al
-
to
-
n
o
is
e
r
atio
Co
p
y
rig
h
t
©
2
0
2
0
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
:
C
h
a
n
i
n
to
r
n
J
itta
w
ir
i
y
an
u
k
o
o
n
,
Gr
ad
u
ate
Sch
o
o
l o
f
A
d
v
a
n
ce
d
T
ec
h
n
o
lo
g
y
Ma
n
a
g
e
m
e
n
t,
Ass
u
m
p
t
io
n
U
n
iv
er
s
it
y
,
Sa
m
u
t P
r
ak
an
P
r
o
v
in
ce
,
1
0
5
4
0
,
T
h
ailan
d
.
E
m
ail: p
ct2
5
2
6
@
y
a
h
o
o
.
co
m
1.
I
NT
RO
D
UCT
I
O
N
T
h
e
m
is
s
in
g
d
ata
ar
e
p
er
v
a
s
i
v
e
in
t
h
e
ca
lc
u
lati
n
g
p
r
ac
tice.
T
h
ey
ca
n
b
e
m
i
s
s
i
n
g
f
o
r
s
o
m
e
in
s
tan
ce
s
o
r
attr
ib
u
tes
[1
-
3]
.
I
f
m
ai
n
s
tr
e
a
m
(
b
u
l
k
)
o
f
d
ata
i
s
m
i
s
s
i
n
g
o
n
t
h
e
at
tr
ib
u
te
t
h
e
n
it
is
alle
g
e
d
to
b
e
u
n
n
o
ticed
.
T
r
a
d
itio
n
al
tr
ea
t
m
en
t
s
a
n
d
s
o
f
t
w
ar
e
al
w
a
y
s
a
s
s
u
m
e
t
h
at
all
a
ttrib
u
tes
in
a
d
ata
s
et
ar
e
f
i
g
u
r
ed
f
o
r
all
in
s
ta
n
ce
s
.
T
h
e
p
o
p
u
lar
m
et
h
o
d
f
o
r
all
f
u
n
d
a
m
en
tal
s
o
f
t
w
ar
e
i
s
to
eli
m
i
n
ate
i
n
s
ta
n
ce
s
w
it
h
a
n
y
n
o
is
e
a
tec
h
n
iq
u
e
i
s
k
n
o
w
n
a
s
co
m
p
lete
d
ata
a
n
al
y
tics
[4
-
6]
.
T
h
e
ev
id
e
n
t
w
ea
k
n
es
s
o
f
eli
m
atio
n
is
th
a
t
i
n
ca
s
e
o
f
b
u
l
k
n
o
is
e,
it
h
ab
itu
al
l
y
ca
n
ce
l
s
a
h
e
f
t
y
p
o
r
tio
n
o
f
th
e
attr
ib
u
te,
r
e
s
u
l
tin
g
to
a
b
o
ld
lo
s
s
o
f
n
u
m
er
ica
l
i
m
p
lica
tio
n
.
Da
ta
s
cien
t
is
t
is
p
la
u
s
ib
l
y
u
n
w
ill
in
g
to
ab
an
d
o
n
d
ata
h
e
h
as
s
p
e
n
t
m
o
n
e
y
,
e
f
f
o
r
t,
an
d
ti
m
e
in
ac
cu
m
u
lati
n
g
.
As
s
u
c
h
,
m
o
s
t tr
ea
t
m
e
n
t te
c
h
n
iq
u
es f
o
r
th
e
ca
s
e
w
i
th
b
u
lk
n
o
is
e
h
av
e
b
ec
o
m
e
p
r
o
m
i
n
en
t.
P
am
p
ak
a,
et
al
[
7
]
d
ef
in
e
m
i
s
s
i
n
g
v
al
u
es
as
t
h
e
n
o
is
e
w
h
i
ch
is
n
o
t
d
ep
o
s
ited
f
o
r
an
en
tit
y
i
n
th
e
in
s
ta
n
ce
o
f
in
ter
est.
T
h
e
co
m
p
licatio
n
o
f
m
i
s
s
i
n
g
v
alu
e
is
co
r
p
o
r
ate
in
m
o
s
t
r
esear
c
h
es
a
n
d
r
ef
lects
n
o
n
tr
i
v
ial
co
n
clu
s
io
n
s
.
Ma
n
y
t
y
p
es
o
f
r
esear
ch
h
a
v
e
atten
d
ed
to
tr
e
at
th
e
n
o
is
e
an
d
p
r
o
b
l
em
s
ar
is
en
f
r
o
m
m
i
s
s
i
n
g
v
alu
e
s
,
a
n
d
th
e
ap
p
r
o
ac
h
es
t
o
p
r
ev
en
t
p
ar
ticu
lar
l
y
i
n
t
h
e
m
ed
ical
ar
ea
[
8
]
.
Dziu
r
a,
et
al
[
9
]
in
tr
o
d
u
ce
th
e
p
r
o
m
i
s
in
g
ap
p
r
o
ac
h
o
f
tr
ea
t
in
g
t
h
e
n
o
is
e
i
s
to
av
o
id
t
h
e
i
s
s
u
e
b
y
w
ell
-
d
esi
g
n
th
e
s
t
u
d
y
a
n
d
a
m
as
s
in
g
t
h
e
d
ata
p
r
u
d
en
tl
y
.
Ma
l
l
in
c
k
r
o
d
t,
et
al
[
10
]
a
r
e
s
ig
n
i
f
y
in
g
to
les
s
en
th
e
a
m
o
u
n
t
o
f
n
o
i
s
e
in
t
h
e
s
ci
en
ti
f
ic
s
t
u
d
y
.
T
h
e
y
p
r
o
p
o
s
e
th
e
p
lan
n
i
n
g
h
as
to
ed
g
e
th
e
d
ata
ac
c
u
m
u
latio
n
to
r
e
s
ea
r
ch
er
s
.
T
h
is
ca
n
b
e
attain
e
d
b
y
d
ec
r
ea
s
i
n
g
t
h
e
n
u
m
b
er
o
f
cr
itical
d
ata
co
llectin
g
,
in
v
esti
g
atio
n
s
,
an
d
u
s
i
n
g
th
e
b
ef
itti
n
g
v
i
s
u
al
izatio
n
.
P
r
io
r
to
th
e
s
tu
d
y
,
a
co
m
p
r
e
h
en
s
iv
e
d
o
cu
m
en
tatio
n
o
f
t
h
e
r
esear
c
h
i
s
to
p
r
ep
ar
e
th
e
g
u
id
e
o
f
o
p
er
atio
n
s
i
n
c
lu
d
in
g
th
e
w
a
y
s
to
s
elec
t
t
h
e
m
e
m
b
er
s
,
p
r
o
ce
d
u
r
e
to
tr
ain
t
h
e
m
e
m
b
er
s
,
th
e
n
o
is
e
tr
ea
t
m
e
n
t,
as
w
ell
a
s
p
r
o
c
ess
to
co
llect
a
n
d
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
5
0
2
-
4752
I
n
d
o
n
esia
n
J
E
lec
E
n
g
&
C
o
m
p
Sci,
Vo
l.
17
,
No
.
1
,
J
an
u
ar
y
20
20
:
5
4
3
-
550
544
r
ev
is
e
t
h
e
d
ata.
B
esid
es,
if
a
tr
iv
ial
p
r
o
j
ec
t
is
tar
g
eted
b
ef
o
r
e
th
e
p
r
im
ar
y
co
llectio
n
,
it
m
a
y
h
elp
d
etec
t
th
e
u
n
p
r
ed
ictab
le
co
m
p
licatio
n
s
w
h
ic
h
m
a
y
ar
is
e
d
u
r
i
n
g
th
e
r
esear
ch
,
as
w
ell
as
s
i
n
k
i
n
g
th
e
n
u
m
b
er
o
f
m
is
s
i
n
g
v
al
u
es.
I
n
r
ep
eti
tio
n
,
b
u
lk
n
o
i
s
e
[
11
]
cu
lti
v
ates
w
h
en
e
v
er
u
n
r
ec
o
g
n
iz
ed
ch
ar
ac
ter
s
i
n
cl
u
d
in
g
n
u
ll,
b
lan
k
,
an
d
o
th
er
s
h
av
e
o
cc
u
p
ied
an
y
r
o
w
s
as
s
h
o
w
n
in
t
h
e
tab
le
o
f
Fig
u
r
e
1
.
No
is
e
d
ata
ca
n
c
u
lti
v
ate
a
n
er
r
atic
co
n
s
eq
u
e
n
ce
v
ar
y
i
n
g
f
r
o
m
t
h
e
er
r
o
n
eo
u
s
d
ataset
to
n
o
n
r
esp
o
n
s
iv
e
ex
ec
u
tio
n
.
Vir
m
a
n
i,
et
al
[
12
]
in
tr
o
d
u
ce
a
clu
s
ter
i
n
g
al
g
o
r
ith
m
b
a
s
ed
o
n
K
-
m
ea
n
s
i
n
o
r
d
er
to
r
all
y
r
es
u
lts
f
o
r
u
s
er
s
o
v
er
s
o
cial
n
e
t
wo
r
k
s
.
T
h
e
K
-
m
ea
n
s
alg
o
r
ith
m
p
er
s
e
allo
w
s
th
e
r
e
s
ea
r
ch
er
to
f
i
x
t
h
e
K
v
alu
e.
T
h
e
p
ap
er
b
ased
o
n
th
e
f
ix
ed
f
i
g
u
r
e
o
f
K
i
m
p
r
o
v
e
s
7
0
%
in
s
i
m
i
lar
it
y
ex
p
er
i
m
en
t.
Sh
i,
et
al
[
13
]
in
v
e
s
tig
a
te
an
i
n
n
o
v
ati
v
e
alg
o
r
it
h
m
to
o
p
t
th
e
f
it
n
ess
ca
lcu
lat
io
n
to
th
e
u
n
io
n
f
u
n
ctio
n
in
K
-
Me
an
s
al
g
o
r
ith
m
.
R
e
s
u
lt
s
b
ased
u
p
o
n
th
e
co
m
b
i
n
atio
n
o
f
th
e
s
e
f
u
n
ctio
n
s
af
f
o
r
d
a
b
etter
co
m
p
r
eh
e
n
s
i
v
e
d
o
cu
m
en
t.
W
ar
tan
a,
et
al
[
14
]
in
tr
o
d
u
ce
a
Fu
zz
y
-
b
ased
al
g
o
r
it
h
m
to
in
cr
ea
s
e
t
h
e
s
ec
u
r
it
y
a
n
d
s
tab
ilit
y
o
f
t
h
e
p
o
w
er
s
y
s
te
m
.
I
t
p
r
o
v
es
t
h
at
th
e
f
u
zz
y
al
g
o
r
ith
m
i
s
s
u
p
p
o
r
tin
g
t
h
e
d
ec
is
io
n
m
ak
in
g
m
o
r
e
ef
f
ec
ti
v
el
y
t
h
a
n
th
e
g
e
n
etic
alg
o
r
it
h
m
.
Ma
n
o
j
et
al
[
15
]
p
r
o
p
o
s
e
th
e
p
r
e
d
ictiv
e
f
r
a
m
e
w
o
r
k
b
ased
o
n
th
e
n
e
u
r
al
n
e
t
w
o
r
k
m
o
d
e
l
f
o
r
o
p
ti
m
al
p
er
f
o
r
m
an
ce
o
f
t
h
e
r
eu
s
ab
ilit
y
o
f
t
h
e
co
d
e
.
T
h
e
least
s
q
u
ar
e
alg
o
r
ith
m
also
i
s
u
s
ed
to
o
b
tain
o
p
ti
m
izatio
n
in
o
r
d
er
to
ca
lcu
late
an
d
co
n
f
ir
m
th
e
h
i
g
h
e
s
t r
eliab
ilit
y
.
B
u
lk
n
o
is
e
r
ep
r
esen
ts
a
n
y
u
n
r
ea
d
ab
le
an
d
u
s
eles
s
d
ata
w
h
ic
h
is
co
llected
u
n
in
ten
t
i
o
n
all
y
,
b
u
t
o
b
s
cu
r
es.
Su
r
es
h
et
al
[
1
6
]
tr
ea
t
a
d
en
o
is
ed
p
r
o
ce
s
s
to
im
p
r
o
v
e
th
e
s
p
ec
tr
al
o
f
s
atelli
te
i
m
a
g
e.
T
h
ese
Gau
s
s
ia
n
n
o
is
es
ar
e
co
n
ta
m
i
n
ati
n
g
n
o
t
o
n
l
y
co
r
r
u
p
ted
p
r
o
b
le
m
s
s
u
c
h
as
h
ar
d
w
ar
e
o
r
s
o
f
t
w
ar
e
in
co
m
p
a
tib
ilit
y
b
u
t
also
p
r
o
ce
s
s
in
g
v
u
ln
er
ab
ilit
ie
s
s
u
c
h
as
n
o
f
u
r
th
er
e
x
ec
u
t
io
n
,
o
r
n
o
o
p
er
atio
n
,
o
r
f
ailu
r
e.
A
b
u
l
k
n
o
is
e
ca
n
r
u
i
n
th
e
class
i
f
y
in
g
p
r
o
ce
s
s
o
f
t
h
e
d
a
taset.
I
n
t
h
is
ca
s
e,
b
u
lk
n
o
is
e
w
o
r
s
e
n
s
t
h
e
s
tab
ili
t
y
a
n
al
y
s
is
a
n
d
r
e
m
ai
n
s
a
n
ex
ce
s
s
iv
e
r
is
k
.
T
o
d
en
o
is
e
s
at
ellite
i
m
a
g
es
is
cr
itical
f
o
r
i
m
p
r
o
v
in
g
t
h
e
v
is
u
aliza
t
io
n
o
f
i
m
ag
e
s
a
n
d
f
o
r
ea
s
i
n
g
s
u
p
p
le
m
e
n
tar
y
a
n
al
y
s
is
an
d
it
s
p
r
o
ce
s
s
in
g
ta
s
k
s
.
T
o
t
a
l
A
mo
u
n
t
U
n
i
t
1
U
n
i
t
2
U
n
i
t
3
U
n
i
t
4
U
n
i
t
5
T
r
a
n
s A
1
0
2
3
4
XX
*
&
^
T
r
a
n
s B
T
r
a
n
s C
2
3
4
.
6
CH
9
0
7
6
!!!
T
r
a
n
s D
A
Z
X
T
r
a
n
s E
3
4
2
.
4
6
@#
N
/
A
Fig
u
r
e
1
.
An
E
x
a
m
p
le
o
f
No
is
e
P
atter
n
.
B
lan
k
I
n
d
icate
s
t
h
at
t
h
e
Val
u
e
i
s
Mi
s
s
i
n
g
T
h
e
o
b
j
ec
tiv
e
o
f
t
h
e
r
esear
c
h
is
to
i
n
v
e
s
ti
g
ate
t
h
e
ac
cu
r
ac
y
o
f
th
e
r
eg
r
es
s
io
n
m
o
d
el
f
o
r
b
u
lk
n
o
i
s
e
d
ata
u
s
in
g
MO
A
[
1
7
]
.
I
n
th
e
an
al
y
s
is
,
a
lar
g
e
p
o
r
tio
n
o
f
n
o
is
e
is
f
o
u
n
d
to
b
e
ab
o
v
e
f
if
t
y
p
er
ce
n
t
o
f
th
e
to
tal
s
ize
o
f
th
e
d
ataset.
T
h
is
i
s
ca
ll
ed
,
"
b
u
lk
n
o
is
e"
w
h
ic
h
is
illo
g
ical
f
l
u
ctu
a
tio
n
d
u
e
to
attr
ib
u
t
e
w
h
ich
i
s
n
o
t
ab
le
to
b
e
ac
co
u
n
ted
f
o
r
.
B
u
lk
n
o
is
e
w
ill
b
e
co
n
s
id
er
ed
f
r
o
m
p
r
ac
tical
p
o
in
ts
o
f
v
ie
w
.
T
h
e
n
o
is
e
p
ar
t
th
u
s
n
ee
d
s
to
b
e
d
etec
ted
in
o
r
d
er
to
b
r
ea
k
th
r
o
u
g
h
t
h
e
f
ail
u
r
e
in
m
a
n
ip
u
latio
n
.
Ne
x
t,
t
h
e
p
r
o
p
o
s
ed
alg
o
r
ith
m
w
ill
tr
ea
t
th
ese
n
o
i
s
es
th
e
n
p
r
ed
ictio
n
r
esu
lt
s
f
r
o
m
s
i
m
u
latio
n
ar
e
co
ll
ec
ted
to
leg
al
ize
t
h
e
ac
c
u
r
ac
y
.
Fin
a
ll
y
,
t
h
e
co
r
r
ec
tn
ess
o
f
t
h
e
p
r
o
p
o
s
ed
tr
e
at
m
e
n
t
w
ill b
e
co
m
p
ar
ed
w
it
h
th
e
ac
tu
a
l d
ata.
2.
RE
L
AT
E
D
WO
RK
C
o
n
s
er
v
ati
v
e
s
tati
s
tical
co
m
p
u
tatio
n
a
n
d
s
o
f
t
w
ar
e
co
u
n
t
o
n
co
llected
in
s
ta
n
ce
s
i
n
an
in
d
icate
d
f
r
a
m
e
w
o
r
k
f
o
r
en
tire
ca
s
e
s
.
Fo
r
a
len
g
t
h
y
ti
m
e,
t
h
e
m
is
s
i
n
g
d
ata
is
ex
p
lai
n
ed
as
th
e
„
u
n
k
n
o
w
n
‟
o
f
co
m
p
u
tatio
n
.
A
lt
h
o
u
g
h
m
o
s
t
ca
s
es
ex
p
er
ien
ce
m
i
s
s
i
n
g
v
alu
e
an
d
r
eq
u
ir
e
tr
ea
tin
g
t
h
e
p
r
o
b
lem
i
n
s
o
m
e
tech
n
iq
u
es,
t
h
er
e
i
s
ab
s
o
lu
te
l
y
n
o
th
i
n
g
f
o
u
n
d
in
t
h
e
l
iter
atu
r
e
o
r
p
r
ac
tical
g
u
id
an
ce
.
I
t
i
s
s
o
f
ar
b
ec
au
s
e
n
o
n
e
o
f
th
e
w
id
el
y
u
s
ed
m
e
th
o
d
s
h
av
e
an
y
co
n
cr
ete
ca
lcu
latio
n
s
.
A
m
e
th
o
d
f
o
r
d
ea
lin
g
w
i
th
t
h
e
m
is
s
in
g
v
al
u
e
s
is
p
r
esen
ted
[
1
8
]
as
th
e
tem
p
o
r
a
l
d
ata
is
u
n
s
u
r
p
r
is
in
g
l
y
r
ec
u
r
r
in
g
u
s
i
n
g
d
i
f
f
er
e
n
t
d
is
cr
etiza
ti
o
n
tech
n
iq
u
es.
T
h
e
co
n
ce
p
t
o
f
e
x
cl
u
s
io
n
o
r
in
cl
u
s
io
n
o
f
:
a
te
m
p
o
r
al
s
eq
u
e
n
ce
o
f
t
h
e
d
ata,
clas
s
i
f
icatio
n
lab
el,
an
d
m
an
a
g
i
n
g
o
f
s
tr
ea
m
d
ata
f
o
r
te
m
p
o
r
al
d
ata
d
is
cr
etiza
tio
n
is
ap
p
lied
.
T
h
e
p
r
er
eq
u
is
ite
is
th
a
t
d
ata
n
ee
d
s
to
p
er
s
is
t.
T
h
e
au
th
o
r
s
[
1
9
]
p
r
esen
t
th
e
r
eg
r
e
s
s
io
n
m
o
d
els
w
h
e
r
e
t
h
e
p
r
i
m
a
r
y
r
elatio
n
s
h
ip
e
m
b
r
ac
es
i
n
ter
ac
tio
n
ex
p
r
ess
io
n
s
.
A
lin
ea
r
f
r
a
m
e
w
o
r
k
w
ith
o
n
e
f
u
ll
y
w
it
n
e
s
s
ed
p
r
ed
icto
r
is
co
n
s
id
er
ed
.
T
h
en
th
e
co
n
d
iti
o
n
al
d
is
tr
ib
u
tio
n
o
f
in
ter
ac
tio
n
ex
p
r
es
s
io
n
an
d
th
e
m
i
s
s
i
n
g
co
v
ar
ian
ce
is
ap
p
lied
f
o
r
e
x
a
m
in
i
n
g
t
h
e
p
er
f
o
r
m
an
ce
o
f
m
u
l
tip
le
i
m
p
u
ta
tio
n
s
.
Oth
er
tech
n
iq
u
es
w
h
ic
h
ca
n
b
e
e
m
p
lo
y
ed
b
y
ad
j
u
s
tin
g
m
u
l
tip
le
i
m
p
u
tatio
n
s
o
f
t
w
ar
e
to
o
u
tp
er
f
o
r
m
in
s
p
ite
o
f
i
n
c
o
m
p
atib
il
ities
b
et
w
ee
n
u
n
d
er
l
y
in
g
r
elatio
n
s
h
ip
s
a
m
o
n
g
th
e
attr
ib
u
te
s
an
d
f
r
a
m
e
w
o
r
k
ass
u
m
p
tio
n
s
ar
e
in
v
esti
g
ated
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
d
o
n
esia
n
J
E
lec
E
n
g
&
C
o
m
p
Sci
I
SS
N:
2502
-
4752
P
r
o
p
o
s
ed
a
lg
o
r
ith
m
fo
r
R
eg
r
ess
io
n
-
b
a
s
ed
p
r
ed
ictio
n
w
ith
b
u
lk
n
o
is
e
(
C
h
a
n
in
to
r
n
Jitta
w
ir
iya
n
u
ko
o
n
)
545
No
n
eth
ele
s
s
,
th
e
ex
p
er
i
m
e
n
t
i
n
t
h
is
r
esear
ch
d
o
es
n
o
t
s
h
ad
o
w
an
y
ap
p
r
o
ac
h
es
as
m
e
n
tio
n
ed
ea
r
lier
.
T
h
e
p
r
o
p
o
s
ed
tr
ea
tm
e
n
t
b
eg
in
s
w
it
h
t
h
e
u
n
w
a
n
ted
b
u
lk
n
o
is
e
class
i
f
ica
tio
n
.
A
f
ter
th
at,
th
e
p
r
o
p
o
s
ed
alg
o
r
ith
m
r
ep
air
s
all
u
n
w
a
n
t
ed
ele
m
en
ts
i
n
t
h
e
d
ataset
b
y
o
b
tai
n
i
n
g
a
lo
ca
l
o
p
ti
m
a
l
s
o
lu
tio
n
at
ea
c
h
co
m
p
u
ti
n
g
s
tep
a
n
d
c
h
o
o
s
es
t
h
e
b
est
s
o
lu
tio
n
in
v
ie
w
o
f
g
l
o
b
al
i
m
p
ac
ts
.
No
te
th
at
ev
e
n
i
f
t
h
e
s
in
g
le
ele
m
en
t
o
f
n
o
is
e
in
t
h
e
d
ataset
ca
n
i
m
p
ed
e
t
h
e
d
ata
r
u
n
n
in
g
u
n
le
s
s
th
e
e
x
clu
s
io
n
o
f
t
h
e
n
o
i
s
e.
T
h
e
ex
is
ti
n
g
t
w
o
alg
o
r
ith
m
s
,
n
a
m
el
y
,
Me
an
Va
r
iab
les
(
MV
)
,
an
d
R
an
d
o
m
I
m
p
u
tatio
n
(
R
I
)
ar
e
ap
p
lied
f
o
r
r
ep
air
in
g
n
o
is
e
w
i
t
h
s
u
b
s
t
itu
t
io
n
.
T
h
u
s
,
th
e
co
m
p
u
tatio
n
co
s
t
s
w
h
ich
ar
e
i
n
cl
u
s
i
v
e
o
f
s
ea
r
ch
i
n
g
ti
m
e
f
o
r
b
u
lk
n
o
is
e
r
e
m
o
v
al
a
n
d
alg
o
r
ith
m
r
u
n
ti
m
e
w
ill
b
e
cited
.
T
h
ese
t
w
o
al
g
o
r
ith
m
s
a
r
e
co
m
p
ar
ed
w
it
h
ac
tu
al
v
al
u
es
to
r
ef
lect
th
e
ir
p
r
ec
is
io
n
s
.
T
h
e
ex
p
er
i
m
en
ta
l
r
esu
lt
s
u
s
i
n
g
MO
A
s
i
m
u
la
ti
o
n
ar
e
co
llected
to
ch
ec
k
t
h
e
ac
cu
r
ac
y
b
et
w
ee
n
ex
is
t
in
g
an
d
p
r
o
p
o
s
ed
alg
o
r
ith
m
s
.
T
h
e
a
w
ait
in
g
o
u
tli
n
e
o
f
t
h
e
r
esear
ch
is
as
f
o
llo
w
s
.
I
n
s
ec
ti
o
n
I
I
I
,
b
u
lk
n
o
is
e
co
n
d
itio
n
s
ar
e
in
tr
o
d
u
ce
d
.
S
ec
tio
n
I
V
ex
p
lain
s
th
e
p
er
f
o
r
m
an
ce
r
es
u
lt
s
o
f
th
e
p
r
o
p
o
s
ed
alg
o
r
ith
m
f
r
o
m
ex
p
er
i
m
e
n
tal
p
er
s
p
ec
tiv
e.
Sect
io
n
V
f
i
n
all
y
o
u
tl
in
e
s
th
e
co
n
c
lu
s
io
n
o
f
t
h
e
r
esear
ch
.
3.
B
UL
K
NO
I
SE
CH
A
RAC
T
E
RIS
T
I
C
S
C
h
ar
ac
ter
is
tics
o
f
b
u
l
k
m
is
s
i
n
g
v
a
lu
e
s
ar
e
d
is
cu
s
s
ed
,
d
atase
ts
w
it
h
b
u
l
k
n
o
is
e
ar
e
ill
u
s
tr
at
ed
in
th
is
s
ec
tio
n
.
No
te
t
h
at
a
f
e
w
e
n
tr
i
es
o
f
n
o
i
s
e
ca
n
cr
o
o
k
a
d
atas
et
as
th
e
w
h
o
le.
B
u
l
k
n
o
is
e
ca
n
d
ev
e
lo
p
m
u
c
h
h
ig
h
er
i
m
p
ac
t
th
a
n
e
v
er
a
s
it
ca
n
ce
r
tai
n
l
y
cr
ea
te
f
a
u
lts
d
u
r
in
g
d
ata
co
m
p
i
lin
g
o
r
s
to
r
in
g
.
A
n
o
is
e
b
lo
ck
s
t
h
e
in
s
i
g
h
t
e
x
tr
ac
tio
n
i
n
d
ata
cu
r
atio
n
,
w
h
ic
h
ca
n
r
esu
lt
i
n
t
h
e
ab
o
r
ted
d
ee
p
lear
n
in
g
o
p
er
atio
n
.
I
t
ca
n
b
e
f
r
an
tica
ll
y
co
m
p
le
x
to
lev
er
a
g
e
th
e
f
a
u
lts
.
As
s
u
c
h
,
to
class
i
f
y
a
n
d
tr
ea
t
th
e
n
o
i
s
e
d
ata
ar
e
a
m
u
s
t
to
o
v
er
co
m
e
th
e
co
n
s
tr
ai
n
t.
I
n
th
i
s
r
esear
ch
,
th
e
o
v
er
w
h
el
m
ca
s
e
o
f
n
o
is
e
in
th
e
d
ataset
is
s
t
u
d
ied
.
B
u
lk
n
o
is
e
r
ev
e
n
u
e
s
th
e
atten
d
an
ce
o
f
n
o
is
e
in
t
h
e
d
at
aset
to
b
e
o
u
t
s
id
e
5
0
%.
T
h
e
co
n
v
o
lu
tio
n
is
to
q
u
est
s
y
s
te
m
atica
ll
y
w
h
er
e
t
h
e
b
u
lk
n
o
i
s
e
ac
co
m
p
an
ie
s
.
T
h
e
s
ea
r
ch
co
n
cl
u
d
es
th
e
ess
e
n
ce
o
f
th
e
b
id
o
f
n
o
is
e
tr
ea
t
m
en
t.
T
o
ter
m
in
ate
b
u
l
k
n
o
is
e,
t
h
e
d
eter
m
i
n
is
t
ic
d
atase
t
at
h
a
n
d
f
o
r
ex
ec
u
tio
n
i
s
as
s
u
m
ed
.
I
n
th
is
r
esear
c
h
,
a
s
p
lit
-
an
d
-
r
ep
air
is
ta
k
en
o
n
b
y
e
x
p
ec
tin
g
th
at
a
d
ataset
D
ca
n
b
e
s
p
lit
in
to
t
w
o
p
ar
ts
:
a
m
i
n
o
r
b
u
t
clea
n
p
ar
t,
Dc
an
d
a
b
u
lk
n
o
is
e
p
ar
t,
Dn
.
I
n
t
h
e
n
o
i
s
y
e
n
v
ir
o
n
m
en
t
(
Dn
≥
Dc)
,
th
e
ass
u
m
p
tio
n
i
s
m
o
r
e
r
ep
r
esen
tati
v
e.
Ho
w
e
v
er
,
in
ca
s
e
o
f
th
e
g
ig
a
n
tic
d
ata
s
et,
to
p
u
r
g
e
b
u
lk
n
o
is
e
i
s
asce
n
d
in
g
u
p
t
h
e
s
p
lit
-
an
d
-
r
ep
air
ti
m
e
co
r
r
esp
o
n
d
in
g
l
y
.
T
h
e
s
i
m
u
lat
io
n
o
n
t
h
e
d
ataset
w
it
h
b
u
lk
n
o
is
e
d
is
p
la
y
s
t
h
e
s
u
f
f
ici
en
t p
er
f
o
r
m
a
n
ce
ac
co
r
d
in
g
l
y
.
A
g
e
n
er
al
ap
p
r
o
ac
h
to
d
ea
l
with
b
u
lk
n
o
i
s
e
d
ata
is
to
p
u
r
g
e
all
i
n
s
ta
n
ce
s
co
n
tai
n
i
n
g
th
e
n
o
is
e.
B
u
t,
th
e
tec
h
n
iq
u
e
as
s
u
ch
w
ill
n
o
t
ir
o
n
o
u
t
t
h
e
b
u
lk
n
o
i
s
e
p
r
o
b
lem
as,
o
n
l
y
a
Dc
r
e
m
ain
s
.
No
t
to
m
e
n
tio
n
,
r
e
m
o
v
ed
i
n
s
ta
n
ce
s
ca
n
a
f
f
ec
t t
h
e
o
n
g
o
i
n
g
d
ata
c
u
r
atio
n
.
T
o
s
cr
ee
n
D
n
i
n
t
h
e
d
ata
s
et,
t
h
e
e
x
is
ten
t b
o
u
n
d
o
f
th
e
n
o
is
e
is
p
r
es
u
m
ed
.
T
h
en
,
o
p
ti
m
izatio
n
is
p
r
o
b
ab
le
o
n
th
e
s
i
m
u
latio
n
.
T
h
e
s
p
lit
-
a
n
d
-
r
ep
air
m
eth
o
d
f
o
r
Dn
is
a
m
ain
ta
r
g
et
o
f
t
h
e
r
esear
ch
as
b
u
l
k
n
o
is
e
u
n
les
s
p
u
r
g
i
n
g
ca
n
d
is
co
n
ti
n
u
e
f
u
r
t
h
er
d
ata
an
aly
tics
.
T
w
o
ap
p
r
o
ac
h
es
f
o
r
esti
m
ati
n
g
d
ata
f
o
r
Dn
w
h
ic
h
ar
e
Me
an
Var
iab
les
(
MV
)
,
an
d
R
an
d
o
m
I
m
p
u
ta
tio
n
(
R
I
)
h
av
e
b
ee
n
i
n
tr
o
d
u
ce
d
.
L
et
D
b
e
a
d
ataset
m
atr
i
x
w
h
i
ch
co
n
tai
n
s
a
r
o
w
s
an
d
b
co
lu
m
n
s
,
w
h
ile
n
r
ep
r
esen
t
s
in
s
tan
ce
s
af
f
ec
ted
b
y
n
o
is
e,
in
w
h
ich
n
is
al
w
a
y
s
les
s
th
an
a
(
n
<
a
an
d
Dn
1
,
Dn
2
,
Dn
3
,
…,
Dn
(
b
-
1
)
,
Dn
b
)
f
o
r
ea
ch
n
=
1
,
2
,
3
,
…,
a.
T
h
e
D
m
atr
ix
i
s
ex
p
ec
ted
t
o
b
e
a
d
eter
m
i
n
is
tic
s
et.
A
n
ele
m
e
n
t
D
n
b
is
s
et
o
f
th
e
n
o
i
s
y
ele
m
e
n
t
w
h
en
e
v
er
{
Dij
=
ɸ
|
|
∞,
1
≤
i
≤
a;
1
≤
j
≤
b
}.
R
e
m
ar
k
t
h
at
i
n
ca
s
e
o
f
b
u
l
k
n
o
is
e,
n
≥
a/2
.
T
h
e
d
ataset
w
it
h
b
u
lk
n
o
is
e
is
ca
lled
tr
o
u
b
led
d
atase
t.
He
n
ce
,
t
h
e
p
r
o
p
o
s
ed
tr
ea
t
m
e
n
t
to
r
ev
o
l
v
e
t
h
e
h
az
ar
d
an
d
co
n
ti
n
u
e
th
e
an
al
y
s
is
b
y
ap
p
ly
in
g
t
h
e
es
ti
m
at
ed
v
ec
to
r
E
n
i
s
d
escr
ib
ed
in
th
e
n
e
x
t sec
tio
n
.
T
h
e
s
p
lit
-
a
n
d
-
r
ep
air
s
tr
i
k
es
o
u
t
n
o
is
e
w
h
ic
h
ca
n
b
e
s
cr
ee
n
ed
b
y
an
i
m
p
air
ed
f
ilter
in
g
,
b
u
t
eli
m
i
n
ated
in
s
ta
n
ce
s
ca
n
h
a
m
p
er
th
e
an
al
y
tics
.
No
is
e
ca
n
m
i
s
in
ter
p
r
et
to
n
eg
ati
v
e,
in
d
u
ci
n
g
d
ata
s
cien
ce
to
k
ee
p
o
n
w
it
h
f
au
lt
d
ec
is
io
n
(
a
t
y
p
e
o
n
e
er
r
o
r
)
.
I
n
o
r
d
er
to
ass
u
r
e
d
ata
an
aly
s
i
s
,
th
ese
D
k
b
m
u
s
t
b
e
d
ef
in
i
tel
y
d
en
o
is
ed
.
I
t
is
cr
u
cial
to
d
etac
h
Dn
,
p
ar
ticu
lar
l
y
f
o
r
th
e
b
u
lk
n
o
is
e
w
h
e
r
e
n
≥
a/2
,
an
y
tech
n
iq
u
es
h
av
e
to
s
tr
es
s
o
n
a
r
e
m
ain
in
g
m
i
n
o
r
f
r
ac
tio
n
o
f
th
e
w
h
o
le
d
ataset.
T
h
is
r
esear
c
h
m
o
ti
v
ates t
h
e
p
r
o
p
o
s
ed
alg
o
r
ith
m
f
o
r
b
u
l
k
n
o
is
e.
T
h
e
s
i
m
u
la
tio
n
is
b
ased
o
n
th
e
r
eg
r
ess
iv
e
m
o
d
el
w
it
h
te
n
s
y
n
th
e
tic
d
atasets
.
I
n
th
e
i
n
d
iv
id
u
al
ex
p
er
i
m
e
n
t,
t
h
e
s
i
m
u
lat
io
n
i
s
r
u
n
f
o
r
th
e
p
r
o
p
o
s
ed
alg
o
r
ith
m
,
Me
a
n
Var
i
ab
les
(
MV
)
an
d
R
an
d
o
m
I
m
p
u
tatio
n
(
R
I
)
af
ter
d
en
o
is
in
g
.
T
h
e
r
esu
lt
s
f
r
o
m
th
r
ee
tr
ea
tm
e
n
t
s
w
ill b
e
co
m
p
ar
ed
to
th
o
s
e
ac
tu
al
d
ata
in
t
h
e
s
u
b
s
eq
u
e
n
t
y
ea
r
.
4.
RE
SU
L
T
S AN
D
AN
AL
Y
SI
S
T
h
e
MO
A
s
i
m
u
latio
n
is
d
esi
g
n
ated
f
o
r
an
al
y
zi
n
g
te
n
d
at
asets
.
T
h
e
i
n
v
e
s
ti
g
atio
n
o
f
a
r
eg
r
ess
io
n
m
o
d
el
f
o
r
b
u
lk
n
o
is
e
le
v
el
(
n
)
is
p
er
f
o
r
m
ed
.
T
h
e
s
tu
d
y
i
s
d
e
p
lo
y
ed
o
n
an
I
n
tel
®
C
o
r
e
™
i5
C
P
U,
1
.
6
0
GHz
P
r
o
ce
s
s
o
r
an
d
8
G
B
R
A
M
o
n
b
o
ar
d
.
T
h
e
d
atasets
ar
e
d
iv
er
s
e
in
f
i
le
s
ize,
in
s
ta
n
ce
s
,
an
d
att
r
ib
u
tes.
4
.
1
.
C
o
rr
ela
t
io
n Co
ef
f
icient
(
CO
E
F
)
T
h
e
C
OE
F
is
o
n
e
o
f
t
h
e
m
e
tr
ics
i
n
t
h
e
s
tat
is
tic
s
.
I
t
is
a
u
s
e
f
u
l
a
n
al
y
s
i
s
w
h
ic
h
ca
lc
u
late
s
th
e
p
o
w
er
co
n
ce
r
n
i
n
g
co
n
n
ec
tio
n
s
an
d
v
ar
iab
les.
I
n
s
tatis
tics
,
th
is
c
o
ef
f
icie
n
t
r
ef
er
s
as
th
e
R
-
tes
t.
I
t
d
ef
in
es
h
o
w
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
5
0
2
-
4752
I
n
d
o
n
esia
n
J
E
lec
E
n
g
&
C
o
m
p
Sci,
Vo
l.
17
,
No
.
1
,
J
an
u
ar
y
20
20
:
5
4
3
-
550
546
p
o
w
er
f
u
l
co
n
n
ec
tio
n
a
m
o
n
g
t
w
o
v
ar
iab
les
i
s
.
T
h
e
f
i
g
u
r
e
r
an
g
es
b
et
w
ee
n
1
.
0
an
d
-
1
.
0
.
I
f
th
e
f
ig
u
r
e
is
n
eg
a
tiv
e
th
en
,
it
d
eter
m
in
e
s
if
o
n
e
d
ec
lin
es,
t
h
e
o
th
er
r
is
es.
A
ls
o
,
i
f
th
e
f
i
g
u
r
e
is
p
o
s
itiv
e,
t
h
en
it
ea
r
n
s
b
o
th
o
f
th
e
m
eith
er
les
s
en
o
r
g
r
o
w
co
llec
tiv
el
y
.
T
h
e
co
m
p
u
tatio
n
f
o
r
th
i
s
m
etr
ic
ca
n
b
e
f
o
u
n
d
i
n
[
20
].
4
.
2
.
M
ea
n Squ
a
re
E
rr
o
r
Me
an
s
q
u
ar
ed
er
r
o
r
(
MSE
)
[
21
]
is
o
n
e
o
f
m
a
n
y
t
y
p
e
s
i
n
s
tati
s
tics
to
en
u
m
er
ate
t
h
e
d
if
f
er
en
ce
s
a
m
o
n
g
t
h
e
s
a
m
p
le
a
n
d
p
o
p
u
la
tio
n
a
w
aited
b
y
a
r
eg
r
es
s
io
n
m
o
d
el.
T
h
e
lo
w
er
t
h
e
MSE
,
th
e
n
ea
r
er
to
th
e
b
est
-
f
it
c
u
r
v
e
is
co
n
c
lu
d
e
d
.
T
h
e
MSE
clar
if
ies
t
h
e
s
tan
d
ar
d
s
tatis
tical
m
etr
ic
o
f
th
e
d
is
s
i
m
ilar
it
y
a
m
o
n
g
o
b
s
er
v
atio
n
an
d
f
o
r
ec
ast.
T
h
e
d
if
f
er
e
n
t
f
ig
u
r
e
is
ca
lc
u
lated
b
y
th
e
tar
g
eted
d
ata
o
v
er
th
e
er
r
o
r
in
th
e
f
o
r
ec
ast.
A
d
ataset
i
n
a
w
o
r
k
i
n
g
s
et
d
r
o
p
s
th
e
er
r
o
r
v
alu
e
f
o
r
th
e
ex
p
er
i
m
en
t
d
ataset.
Fau
lt
r
ate
f
o
r
tr
ain
in
g
d
ataset
w
il
l
b
e
co
m
p
ar
ativ
e
l
y
h
i
g
h
er
t
h
an
t
h
at
o
f
t
h
e
ex
p
er
i
m
en
t set.
I
f
a
n
y
t
w
o
alg
o
r
it
h
m
s
p
r
o
d
u
ce
th
e
l
ik
e
m
ea
n
ab
s
o
l
u
te
er
r
o
r
th
en
MSE
is
d
ep
lo
y
ed
f
o
r
a
d
ec
is
io
n
,
w
h
ich
i
s
t
h
e
o
p
tim
u
m
a
n
s
w
er
.
4
.
3
.
M
ea
n Abs
o
lute
E
rr
o
r
T
h
e
m
ea
n
ab
s
o
l
u
te
er
r
o
r
(
MA
E
)
[
21
]
is
a
f
i
g
u
r
e
d
ep
lo
y
ed
to
ev
alu
a
te
th
e
f
u
s
s
y
f
o
r
ec
asts
.
T
h
e
MA
E
is
an
a
v
er
ag
e
o
f
t
h
e
ab
s
o
lu
te
f
i
g
u
r
e
o
f
f
a
u
lts
a
n
d
ca
n
b
e
d
ef
i
n
ed
as
m
o
d
el
ev
alu
a
tio
n
s
ta
tis
ti
cs.
4
.
4
.
M
ea
n Va
ria
bles
(
M
V)
Me
an
v
al
u
e
cr
iter
io
n
[
22
]
is
t
o
ass
ig
n
d
ata
f
o
r
all
n
in
s
ta
n
ce
s
.
A
p
p
l
y
t
h
e
s
p
lit
-
an
d
-
r
ep
air
to
th
e
D
d
ataset
an
d
clas
s
i
f
y
Dn
,
a
d
ata
s
et
co
m
p
r
is
e
s
o
f
n
in
s
ta
n
ce
s
with
n
o
is
e.
An
y
n
r
o
w
s
o
f
th
e
m
atr
ix
D
p
o
s
s
e
s
s
a
n
ele
m
e
n
t
d
ij
w
it
h
n
o
is
e
d
ata
w
h
er
e
{d
ij
=
ɸ
|
|
∞,
1
≤
i
≤
n
;
1
≤
j
≤
b
}
th
en
th
e
r
o
w
is
s
w
ap
p
ed
b
y
th
e
MV
f
o
r
esti
m
ated
E
n
d
ataset
as li
s
ted
i
n
(
1
):
|
|
∑
(
1
)
T
h
e
in
v
esti
g
atio
n
o
f
th
e
MV
is
th
at
it
is
an
ac
ce
p
tab
le
f
o
r
ec
ast
f
o
r
a
p
a
r
am
eter
o
u
t
o
f
a
n
o
r
m
al
d
is
tr
ib
u
tio
n
.
T
h
is
tr
ea
t
m
e
n
t
s
o
m
e
h
o
w
i
n
d
u
ce
s
a
v
o
lati
le
u
n
f
a
ir
n
es
s
.
No
t
to
m
e
n
tio
n
th
e
MV
is
led
b
y
t
h
e
s
lan
ted
r
ep
lace
m
e
n
t a
s
w
el
l a
s
cu
lti
v
ates t
h
e
s
ize
o
f
s
tate
s
p
a
ce
.
4
.
5
.
R
a
nd
o
m
I
m
p
uta
t
io
n (
RI)
Utilize
s
e
v
er
al
i
m
p
u
tatio
n
s
at
r
an
d
o
m
f
o
r
r
ep
lace
m
en
t.
A
n
alo
g
o
u
s
to
t
h
e
ab
o
v
e
MV
,
th
e
s
p
lit
-
a
n
d
-
r
ep
air
is
ap
p
lied
to
th
e
tar
g
ete
d
D
d
ataset
an
d
r
esu
lts
a
d
ata
s
et
w
it
h
n
i
n
s
ta
n
ce
s
.
An
y
n
r
o
w
s
o
f
t
h
e
m
atr
i
x
D
p
o
s
s
e
s
s
a
n
ele
m
en
t d
ij
w
it
h
n
o
is
e
d
ata
w
h
er
e
{d
ij
=
ɸ
|
|
∞,
1
≤
i ≤
n
; 1
≤
j
≤
b
}
th
en
th
e
r
o
w
is
s
w
i
tch
ed
b
y
t
h
e
R
I
f
o
r
esti
m
ated
E
n
d
ataset.
T
h
e
m
in
i
m
u
m
li
k
eli
h
o
o
d
f
o
u
n
d
in
co
lu
m
n
j
(
w
h
er
e
j
=
1
,
2
,
3
,
…,
b
)
is
m
ar
k
ed
b
y
d
(
m
i
n
)
j
w
h
er
e
d
(
m
i
n
)
j
=
Mi
n
(
d
n
j
)
f
o
r
ea
ch
n
=
1
,
2
,
3
,
…,
(
a
-
n
)
.
L
i
k
e
w
i
s
e,
th
e
m
ax
i
m
u
m
li
k
eli
h
o
o
d
o
f
co
lu
m
n
j
(
w
h
er
e
j
=
1
,
2
,
3
,
…
,
b
)
is
d
ef
in
ed
b
y
d
(
m
a
x
)
j
w
h
er
e
d
(
m
a
x
)
j
=
Ma
x
(
d
n
j
)
f
o
r
ea
c
h
n
=
1
,
2
,
3
,
…,
(
a
-
n
)
.
T
h
e
s
u
b
s
titu
t
io
n
f
o
r
esti
m
ated
E
n
d
ataset
w
it
h
m
u
ltip
le
i
m
p
u
tatio
n
s
f
o
r
n
in
s
ta
n
ce
s
i
n
ea
ch
co
lu
m
n
j
is
r
an
d
o
m
l
y
e
x
p
lai
n
ed
as f
o
llo
ws:
[
]
(
2
)
4
.
6
.
P
ro
po
s
ed
Alg
o
rit
h
m
T
h
e
p
r
o
p
o
s
ed
alg
o
r
ith
m
w
o
r
k
s
s
tr
aig
h
t
f
o
r
w
ar
d
l
y
,
as
d
escr
i
b
ed
in
th
e
f
o
llo
w
i
n
g
s
ta
g
es.
T
h
e
d
ataset
w
il
l
b
e
s
p
lit
i
n
to
Dc
an
d
D
n
.
T
h
e
Dc
p
o
r
tio
n
is
ass
u
m
ed
to
p
r
o
v
id
e
th
e
s
o
lu
t
io
n
.
I
n
g
en
er
al,
it
is
t
h
e
s
p
lit
-
an
d
-
r
ep
air
ap
p
r
o
ac
h
.
T
h
e
s
u
cc
ess
f
u
l
ca
lc
u
lat
io
n
to
co
v
er
u
p
D
n
in
ev
er
y
f
r
ac
tio
n
al
s
tep
i
m
p
o
s
es
o
n
t
h
e
f
r
u
it
f
u
l
ca
lcu
latio
n
o
f
ev
er
y
s
u
b
s
o
lu
t
i
o
n
.
T
h
is
is
ca
lled
th
e
o
p
ti
m
al
f
ea
tu
r
e
s
as
a
n
o
p
ti
m
al
s
o
l
u
tio
n
ca
n
b
e
m
ad
e
o
u
t
o
f
o
p
tim
a
l
s
u
b
s
o
lu
tio
n
s
.
T
o
r
ea
c
h
ac
co
m
p
lis
h
m
e
n
t
at
ea
ch
p
ar
t
ial
s
tep
,
th
e
p
r
o
p
o
s
ed
alg
o
r
ith
m
co
n
te
m
p
lates
t
h
e
s
u
b
s
o
l
u
tio
n
d
ata
o
n
l
y
at
t
h
at
p
ar
tial
s
tep
.
Na
m
el
y
,
th
e
d
ec
i
s
i
o
n
o
f
ea
ch
f
r
ac
tio
n
al
s
tep
th
e
p
r
o
p
o
s
ed
al
g
o
r
ith
m
m
ak
e
s
is
b
ased
o
n
a
g
lo
b
al
co
n
s
eq
u
e
n
ce
.
T
h
is
w
ill
co
m
p
lete
a
g
lo
b
al
p
o
licy
to
o
b
tain
th
e
o
p
ti
m
al
ch
ar
ac
ter
is
tic
a
n
d
is
s
u
f
f
icien
t
to
co
m
p
r
o
m
i
s
e
d
ec
is
i
v
e
g
o
al.
As
a
m
etap
h
o
r
,
it‟
s
a
n
alo
g
o
u
s
to
d
o
in
g
t
h
e
ch
e
s
s
b
y
k
ee
p
in
g
t
h
i
n
k
i
n
g
ah
ea
d
m
o
r
e
th
an
o
n
e
m
o
v
e,
a
n
d
f
i
n
all
y
s
co
r
in
g
t
h
e
g
a
m
e.
T
h
e
p
r
o
p
o
s
ed
alg
o
r
ith
m
n
ee
d
s
n
o
co
m
p
lex
d
ec
i
s
io
n
r
u
le
as
it
o
n
l
y
d
elib
er
ates
all
t
h
e
av
ailab
le
s
u
b
s
o
l
u
tio
n
s
at
ea
c
h
s
tag
e.
T
h
er
e
is
n
o
t
n
ec
es
s
ar
y
to
ca
lcu
la
te
f
ea
s
ib
l
e
d
ec
is
io
n
in
f
er
en
ce
s
t
h
e
n
t
h
e
c
o
m
p
u
tat
io
n
co
s
t
is
ab
o
u
t
O
(
ab
)
.
T
h
e
p
r
o
p
o
s
ed
alg
o
r
ith
m
is
s
u
m
m
ar
ized
in
F
i
g
u
r
e
2
.
State
s
p
ac
e
is
n
o
n
tr
iv
ia
l
to
r
ef
lect
th
e
s
p
ee
d
o
f
co
m
p
u
ti
n
g
co
m
p
le
x
it
y
.
I
n
t
h
is
r
esear
ch
,
t
h
e
co
m
p
u
tatio
n
co
s
t
is
d
er
iv
ed
,
co
r
r
esp
o
n
d
in
g
to
th
e
p
er
f
o
r
m
an
ce
ass
es
s
m
en
t.
I
t
is
d
ec
ep
tiv
e
an
y
f
o
r
ec
ast
s
ar
e
p
r
o
b
lem
atica
l
if
th
e
co
m
p
u
ta
ti
o
n
co
s
t
i
s
ex
tr
ao
r
d
in
ar
y
as
d
e
p
icted
in
T
ab
le
1
.
No
te
th
at
i
n
ca
s
e
o
f
b
u
l
k
n
o
i
s
e,
a
is
al
w
a
y
s
s
m
aller
th
a
n
2
n
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
d
o
n
esia
n
J
E
lec
E
n
g
&
C
o
m
p
Sci
I
SS
N:
2502
-
4752
P
r
o
p
o
s
ed
a
lg
o
r
ith
m
fo
r
R
eg
r
ess
io
n
-
b
a
s
ed
p
r
ed
ictio
n
w
ith
b
u
lk
n
o
is
e
(
C
h
a
n
in
to
r
n
Jitta
w
ir
iya
n
u
ko
o
n
)
547
P
r
o
p
o
se
d
A
l
g
o
r
i
t
h
m
R
e
q
u
i
r
e
:
D
a
t
a
m
a
t
r
i
x
[
D
]
xy
w
i
t
h
x
r
o
w
s a
n
d
y
c
o
l
u
m
n
s
En
s
u
r
e
:
[
D
]
x
y
,
S
=
a
l
l
p
o
t
e
n
t
i
a
l
so
l
u
t
i
o
n
s
i
n
e
a
c
h
c
o
m
p
u
t
a
t
i
o
n
st
e
p
=
{
,
…
,
},
=
c
e
n
t
r
o
i
d
o
f
t
h
e
a
t
t
r
i
b
u
t
e
y
,
=
c
a
n
d
i
d
a
t
e
s i
n
e
a
c
h
c
o
mp
u
t
a
t
i
o
n
s
t
e
p
,
P
r
=
a
p
r
e
mi
u
m so
l
u
t
i
o
n
w
h
e
r
e
P
r
(S
y
)
≥
0
a
n
d
S
y
S
t
f
o
r
I
=
1
t
o
x
do
f
o
r
J
=
1
t
o
y
do
←
0
f
o
r
k
=
1
t
o
S
/
*
*
A
l
l
so
l
u
t
i
o
n
s c
o
m
p
u
t
a
t
i
o
n
*
*
/
F
k
=
a
r
g
/
*
*
S
o
l
u
t
i
o
n
F
k
a
n
d
c
o
r
r
e
sp
o
n
d
i
n
g
C
k
**/
e
n
d
f
o
r
f
o
r
n
=
1
t
o
S
/
*
*
C
h
o
o
se
b
e
st
so
l
u
t
i
o
n
*
*
/
Δ
n
=
|
F
n
-
C
n
|
e
n
d
f
o
r
=
a
r
g
/
*
*
A
n
e
w
b
e
st
f
o
r
t
h
i
s c
o
m
p
u
t
a
t
i
o
n
st
e
p
*
*
/
R
e
t
u
r
n
/
*
*
R
e
g
r
e
ssi
o
n
-
b
a
se
d
c
o
m
p
u
t
a
t
i
o
n
*
*
/
e
n
d
f
o
r
e
n
d
f
o
r
Fig
u
r
e
2
.
P
r
o
p
o
s
ed
alg
o
r
ith
m
T
ab
le
1
.
C
o
m
p
u
ta
tio
n
C
o
m
p
le
x
it
y
o
f
P
r
o
p
o
s
ed
Me
th
o
d
T
r
e
a
t
me
n
t
C
o
mp
u
t
a
t
i
o
n
C
o
mp
l
e
x
i
t
y
MV
O(
ab
)
+
O
(
ab
-
bn
)
≈
O
(
ab
)
RI
O(
ab
)
+
O
(
2
(
ab
-
bn
)
)
≈
O
(
ab
)
P
R
O
P
O
S
ED
O(
ab
)
+
O
(
2
(
ab
-
bn
)
)
≈
O
(
ab
)
I
n
t
h
is
r
e
s
ea
r
ch
,
t
h
e
s
p
lit
-
an
d
-
r
ep
air
s
tr
ateg
y
is
p
r
o
p
o
s
ed
in
o
r
d
er
to
h
a
n
d
le
th
e
b
u
l
k
n
o
is
e.
T
h
e
s
tr
ateg
y
w
i
ll
s
p
lit
a
n
d
r
ep
air
t
h
e
b
u
lk
n
o
i
s
e
p
o
r
tio
n
p
r
io
r
to
th
e
f
o
r
ec
ast.
A
n
o
t
h
er
m
o
d
el
-
b
ased
s
tr
ateg
y
w
ill
r
ath
er
r
ev
ie
w
t
h
e
al
g
o
r
ith
m
p
er
s
e
to
le
v
er
a
g
e
t
h
e
n
o
is
e
b
ef
o
r
e
th
e
u
s
e
o
f
t
h
e
p
ar
a
m
etr
ic
f
o
r
ec
ast.
T
h
e
latter
s
tr
ateg
y
ca
n
b
e
f
o
u
n
d
in
eit
h
e
r
A
NC
OV
A
[
2
3
,
2
4
]
o
r
P
SP
P
ap
p
licatio
n
,
w
h
ic
h
r
elate
s
co
u
n
t
les
s
i
m
p
u
tatio
n
s
f
o
r
in
ter
c
h
a
n
g
i
n
g
t
h
e
n
o
is
e.
W
h
ile
t
h
e
s
p
lit
-
a
n
d
-
r
ep
air
tech
n
iq
u
e
[
2
5
]
g
ea
r
s
p
r
o
s
p
ec
t
d
ata
to
co
n
s
id
er
atio
n
.
T
h
e
m
o
d
el
-
b
ased
alg
o
r
ith
m
i
s
s
o
m
eh
o
w
co
m
p
lex
,
a
n
d
th
e
u
s
er
‟
s
s
k
ill
is
o
b
li
g
ato
r
y
as
it
h
as
b
ee
n
p
r
o
f
o
u
n
d
l
y
d
esig
n
ed
to
r
ep
licate
th
e
p
ar
a
m
etr
ic
o
n
e.
T
h
e
er
r
o
r
v
alu
es
o
f
ten
d
iv
er
g
e
n
t
d
atasets
u
s
i
n
g
MO
A
at
n
o
is
e
v
al
u
e
r
an
g
i
n
g
f
r
o
m
5
0
%
to
8
0
%
ar
e
ex
a
m
i
n
ed
.
T
h
is
is
a
p
r
i
m
i
tiv
e
an
al
y
tics
to
w
ar
d
th
e
n
o
m
i
n
at
ed
d
atasets
,
a
n
d
all
r
esu
lt
s
ar
e
s
h
o
w
n
in
T
ab
le
2
-
5
.
T
h
e
th
r
ee
er
r
o
r
s
in
th
e
tab
l
e
d
is
tin
g
u
is
h
th
e
co
r
r
elatio
n
c
o
ef
f
icie
n
t
(
C
OE
F),
th
e
m
ea
n
s
q
u
ar
ed
er
r
o
r
(
MSE
)
,
an
d
m
ea
n
ab
s
o
lu
te
er
r
o
r
(
MA
E
)
i
n
d
iv
id
u
all
y
.
Data
s
et#
2
g
iv
e
s
lo
w
es
t
f
ig
u
r
e
f
o
r
C
OE
F,
MSE
a
n
d
MA
E
.
T
h
e
r
eg
r
es
s
io
n
-
b
ased
f
o
r
ec
ast i
s
d
ep
icted
in
T
a
b
le
6
.
T
ab
le
2
.
Fo
r
ec
ast
w
i
th
Me
a
n
A
b
s
o
l
u
te
E
r
r
o
r
f
o
r
T
en
Dif
f
er
en
t D
a
taset
s
(
N
=
0
.
5
)
D
a
t
a
se
t
C
O
EF
M
S
E
M
A
E
1
0
.
3
1
1
7
.
2
1
4
.
2
2
0
.
0
8
1
.
8
3
1
.
6
1
3
0
.
2
9
2
8
.
7
2
4
.
9
4
0
.
1
7
3
0
.
2
2
6
.
3
5
0
.
2
8
6
7
.
4
5
7
.
3
6
0
.
7
9
3
.
0
8
2
.
3
7
0
.
1
4
4
9
.
1
3
7
.
9
8
0
.
3
2
1
2
.
7
1
0
.
2
9
0
.
0
4
1
5
.
7
1
3
.
6
10
0
.
2
2
0
.
1
14
T
ab
le
3
.
Fo
r
ec
ast
w
i
th
Me
a
n
A
b
s
o
l
u
te
E
r
r
o
r
f
o
r
T
en
Dif
f
er
en
t D
a
taset
s
(
N
=
0
.
6
)
D
a
t
a
se
t
C
O
EF
M
S
E
M
A
E
1
0
.
2
17
14
2
0
.
0
1
1
.
8
3
1
.
5
4
3
0
.
3
4
2
8
.
7
2
5
.
2
4
0
.
1
6
2
6
.
3
24
5
0
.
3
0
7
1
.
2
6
0
.
9
6
0
.
7
9
3
.
0
7
2
.
2
9
7
0
4
8
.
6
3
7
.
8
8
0
.
3
5
1
2
.
5
10
9
0
.
1
5
1
5
.
6
13
10
0
.
2
4
20
1
4
.
3
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
5
0
2
-
4752
I
n
d
o
n
esia
n
J
E
lec
E
n
g
&
C
o
m
p
Sci,
Vo
l.
17
,
No
.
1
,
J
an
u
ar
y
20
20
:
5
4
3
-
550
548
T
ab
le
4
.
Fo
r
ec
ast
w
i
th
Me
a
n
A
b
s
o
l
u
te
E
r
r
o
r
f
o
r
T
en
Dif
f
er
en
t D
atase
ts
(
N
=
0
.
7
)
D
a
t
a
se
t
C
O
EF
M
S
E
M
A
E
1
0
.
3
17
1
4
.
1
2
0
.
1
8
1
.
8
1
1
.
5
9
3
0
.
3
4
2
8
.
7
2
5
.
2
4
2
8
.
8
0
.
2
8
25
5
0
.
1
9
7
1
.
2
6
1
.
5
6
0
.
7
9
3
.
0
8
2
.
2
9
7
0
.
0
1
4
8
.
9
3
8
.
5
8
0
.
3
5
1
2
.
5
10
9
0
.
0
7
1
4
.
9
1
2
.
9
10
0
.
2
1
2
0
.
1
14
T
ab
le
5
.
Fo
r
ec
ast
w
i
th
Me
a
n
A
b
s
o
l
u
te
E
r
r
o
r
f
o
r
T
en
Dif
f
er
en
t
Data
s
e
ts
(
N
=
0
.
8
)
D
a
t
a
se
t
C
O
EF
M
S
E
M
A
E
1
0
.
3
7
1
7
.
3
5
1
4
.
5
8
2
0
.
1
8
1
.
8
2
1
.
5
9
3
0
.
3
4
2
8
.
7
2
5
.
2
4
0
.
1
7
3
0
.
2
2
6
.
3
5
0
.
0
2
6
9
.
3
5
9
.
4
6
0
.
8
3
2
.
2
3
7
0
.
3
2
4
9
.
5
3
7
.
9
8
0
.
3
5
1
2
.
5
6
10
9
0
.
3
1
1
4
.
5
1
2
.
6
10
0
.
2
1
2
0
.
1
14
T
ab
le
6
.
R
eg
r
ess
io
n
-
B
ased
Fo
r
ec
ast
f
o
r
T
en
Data
s
ets
R
e
g
r
e
ssi
o
n
-
b
a
se
d
F
o
r
e
c
a
st
D
A
TA
S
ET
1
X
9
=
0
.
8
3
8
X
3
+
2
4
.
5
6
2
X
12
=
-
0
.
1
1
7
X
2
+
3
.
8
9
3
X
4
=
1
.
7
8
X
7
+
1
4
7
4
X
1
=
6
.
3
4
X
3
-
6
.
2
X
5
-
5
0
.
3
5
X
6
=
0
.
2
8
X
2
+
0
.
2
3
X
3
+
1
2
8
4
.
3
6
X
7
=
0
.
3
6
X
4
+
0
.
1
8
X
5
+
0
.
2
1
X
6
-
1
8
.
0
9
7
X
1
=
5
.
3
X
4
+
0
.
2
3
X
5
+
1
1
0
.
6
8
X
3
=
-
8
2
.
7
X
2
+
0
.
0
7
X
5
+
4
2
2
.
7
9
X
6
=
-
1
.
2
7
X
2
+
6
2
4
.
5
3
10
X
7
=
-
1
.
4
X
1
+
1
.
3
X
3
-
1
.
4
X
5
-
0
.
4
X
4
-
1
8
.
5
T
ab
les
7
-
1
0
d
is
clo
s
e
an
a
v
er
ag
e
er
r
o
r
f
o
r
t
h
e
r
e
g
r
ess
io
n
-
b
ased
m
o
d
el
a
s
s
o
ciati
n
g
to
th
e
au
t
h
e
n
ti
c
d
ata.
I
n
th
is
r
esear
ch
,
ten
d
is
s
i
m
ilar
d
atasets
ar
e
s
tu
d
ied
at
th
e
d
iv
er
g
en
t
n
o
is
e
lev
e
l
(
n
)
is
ex
ten
d
i
n
g
f
r
o
m
5
0
%
to
8
0
%
as
ch
ar
ted
in
T
ab
le
7
-
1
0
co
r
r
esp
o
n
d
i
n
g
l
y
.
I
n
v
er
y
ca
s
es
o
f
th
e
f
o
r
ec
ast
f
r
o
m
t
h
e
p
r
o
p
o
s
ed
m
et
h
o
d
,
th
e
er
r
o
r
is
lo
w
e
s
t.
Mo
r
eo
v
er
,
in
ca
s
e
o
f
b
u
lk
n
o
is
e,
th
e
co
m
p
u
ta
tio
n
co
m
p
l
ex
it
y
f
o
r
all
th
r
ee
tr
ea
t
m
e
n
ts
is
a
k
i
n
.
I
t
co
n
clu
d
es
th
e
p
r
o
p
o
s
ed
m
et
h
o
d
is
th
e
u
t
m
o
s
t
ef
f
ec
ti
v
e
alg
o
r
ith
m
f
o
r
b
u
l
k
no
is
e
an
a
l
y
t
ics.
T
ab
le
7
.
A
v
er
ag
e
P
er
ce
n
ta
g
e
o
f
E
r
r
o
r
f
o
r
T
en
Dif
f
er
e
n
t D
at
asets
(
N=
0
.
5
)
n
=
0
.
5
D
a
t
a
se
t
MV
RI
P
R
O
P
O
S
ED
1
1
4
.
1
5
1
4
.
1
6
1
3
.
6
7
2
1
6
.
0
3
1
6
.
2
2
1
5
.
8
3
3
1
7
.
6
1
6
.
9
1
5
.
6
6
4
3
5
.
2
7
3
6
.
5
2
4
.
7
5
3
2
.
5
3
5
.
4
7
.
4
6
1
0
.
2
3
1
8
.
1
4
9
.
9
7
5
1
.
4
5
8
.
5
4
1
.
3
8
13
1
3
.
9
1
1
.
8
9
5
4
.
5
62
46
10
1
7
.
7
1
1
7
.
2
3
1
5
.
2
3
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
d
o
n
esia
n
J
E
lec
E
n
g
&
C
o
m
p
Sci
I
SS
N:
2502
-
4752
P
r
o
p
o
s
ed
a
lg
o
r
ith
m
fo
r
R
eg
r
ess
io
n
-
b
a
s
ed
p
r
ed
ictio
n
w
ith
b
u
lk
n
o
is
e
(
C
h
a
n
in
to
r
n
Jitta
w
ir
iya
n
u
ko
o
n
)
549
T
ab
le
8
.
A
v
er
ag
e
P
er
ce
n
ta
g
e
o
f
E
r
r
o
r
f
o
r
T
en
Dif
f
er
e
n
t D
at
asets
(
N=
0
.
6
)
n
=
0
.
6
D
a
t
a
se
t
MV
RI
P
R
O
P
O
S
ED
1
1
4
.
0
8
1
4
.
2
1
1
3
.
6
5
2
1
6
.
0
8
1
6
.
2
8
1
5
.
8
3
3
1
7
.
6
1
6
.
9
1
5
.
6
6
4
3
5
.
2
3
6
.
5
2
4
.
2
5
2
8
.
3
3
0
.
4
6
.
4
6
1
0
.
9
5
2
3
.
1
5
1
0
.
1
7
5
1
.
2
4
9
.
9
3
7
.
5
8
1
3
.
1
1
3
.
6
1
1
.
8
9
5
4
.
9
6
0
.
8
4
7
.
1
10
2
0
.
7
6
2
0
.
7
3
1
8
.
2
9
T
ab
le
9
.
A
v
er
ag
e
P
er
ce
n
ta
g
e
o
f
E
r
r
o
r
f
o
r
T
en
Dif
f
er
e
n
t D
at
asets
(
N=
0
.
7
)
n
=
0
.
7
D
a
t
a
se
t
MV
RI
P
R
O
P
O
S
ED
1
1
4
.
2
2
1
4
.
0
6
1
3
.
5
2
2
1
6
.
0
8
1
6
.
2
1
5
.
7
2
3
1
7
.
6
1
6
.
9
1
5
.
6
6
4
3
5
.
2
3
6
.
5
24
5
2
7
.
4
2
9
.
6
5
.
3
2
6
1
2
.
0
7
2
9
.
4
1
1
.
1
7
51
5
1
.
1
3
9
.
3
8
1
3
.
1
1
4
.
3
1
1
.
9
9
5
2
.
3
5
3
.
1
4
1
.
6
10
2
3
.
5
2
1
.
3
9
1
8
.
9
5
T
ab
le
1
0
.
A
v
er
ag
e
P
er
ce
n
tag
e
o
f
E
r
r
o
r
f
o
r
T
en
Dif
f
er
en
t D
at
asets
(
N=
0
.
8
)
n
=
0
.
8
D
a
t
a
se
t
MV
RI
P
R
O
P
O
S
ED
1
1
3
.
9
8
1
4
.
6
9
1
3
.
4
4
2
1
6
.
0
8
1
5
.
9
7
1
5
.
6
2
3
1
7
.
6
1
6
.
9
1
5
.
6
6
4
3
5
.
2
3
6
.
5
24
5
2
7
.
4
2
8
.
9
4
.
5
2
6
1
1
.
9
3
0
.
2
1
1
.
6
7
5
1
.
4
4
9
.
3
38
8
1
3
.
1
1
3
.
3
1
1
.
8
9
5
5
.
5
5
4
.
4
4
4
.
2
10
2
1
.
0
6
1
8
.
5
1
7
.
6
5.
CO
NCLU
SI
O
N
I
n
th
i
s
p
ap
er
,
co
n
v
en
t
io
n
al
al
g
o
r
ith
m
s
f
o
r
tr
ea
tin
g
n
o
is
e
ar
e
i
m
p
er
f
ec
t.
U
n
d
er
th
e
ce
r
tai
n
co
n
d
itio
n
,
th
e
y
s
er
io
u
s
l
y
h
ar
v
e
s
t
b
o
th
s
t
an
d
ar
d
er
r
o
r
an
d
b
iased
p
ar
a
m
etr
ic
f
o
r
ec
ast.
No
t
to
m
e
n
ti
o
n
,
th
e
co
n
s
er
v
ati
v
e
i
m
p
u
ta
tio
n
s
,
MV
an
d
R
I
m
ec
h
an
i
s
m
s
,
y
iel
d
s
e
v
er
e
av
er
a
g
e
er
r
o
r
f
ig
u
r
es.
T
h
e
p
r
o
p
o
s
ed
m
ec
h
a
n
i
s
m
is
p
r
o
v
e
n
to
b
e
a
b
en
ig
n
ch
o
ice
w
h
en
f
o
r
ec
asti
n
g
r
e
g
r
ess
io
n
m
o
d
els
f
o
r
w
h
ic
h
o
p
ti
m
u
m
s
o
lu
tio
n
i
s
co
n
ce
r
n
ed
.
I
t
also
ex
h
ib
it
s
th
e
b
en
e
f
it
o
f
n
o
t
d
em
an
d
i
n
g
th
e
ex
tr
a
co
m
p
u
tatio
n
co
s
t.
Nex
t
m
o
v
e
w
il
l
in
v
esti
g
ate
o
th
er
d
if
f
er
e
n
t
i
m
p
u
ta
tio
n
s
,
s
o
th
at
t
h
e
s
u
itab
l
e
s
u
b
o
p
ti
m
al
s
o
l
u
tio
n
in
ea
c
h
co
m
p
u
tatio
n
p
h
ase
w
i
ll b
e
f
u
r
t
h
er
in
v
esti
g
ated
.
RE
F
E
R
E
NC
E
S
[1
]
X
.
Zh
u
,
S
.
Zh
a
n
g
,
Z.
Jin
,
a
n
d
Z.
Zh
a
n
g
,
“
M
issin
g
V
a
lu
e
Esti
m
a
ti
o
n
f
o
r
M
ix
e
d
-
A
tt
rib
u
te
Da
ta
S
e
ts
”
,
IEE
E
T
ra
n
sa
c
ti
o
n
s
o
n
Kn
o
wled
g
e
a
n
d
Da
ta
E
n
g
in
e
e
rin
g
,
v
o
l.
2
3
,
n
o
.
1
,
p
p
.
1
1
0
-
1
2
1
,
2
0
1
1
.
[2
]
T
.
De
n
g
,
W
.
F
a
n
,
a
n
d
F
.
G
e
e
rts,
“
Ca
p
tu
rin
g
M
issin
g
T
u
p
les
a
n
d
M
issin
g
V
a
lu
e
s
”,
ACM
T
r
a
n
sa
c
ti
o
n
s
o
n
Da
t
a
b
a
s
e
S
y
ste
ms
,
v
o
l.
4
1
,
issu
e
.
2
,
p
p
.
1
0
:
1
-
1
0
:
4
7
,
2
0
1
6
.
[3
]
M
.
M
.
Ra
h
m
a
n
a
n
d
D.
N.
Da
v
is,
“
M
a
c
h
in
e
L
e
a
rn
in
g
-
Ba
se
d
M
issin
g
V
a
lu
e
Im
p
u
tatio
n
M
e
t
h
o
d
f
o
r
Cli
n
ica
l
Da
tas
e
ts
”
,
IAE
NG T
ra
n
sa
c
ti
o
n
s
o
n
En
g
i
n
e
e
rin
g
T
e
c
h
n
o
lo
g
ies
,
p
p
.
2
4
5
-
2
5
7
,
2
0
1
3
.
[4
]
V
.
B
o
e
v
a
,
L
.
L
u
n
d
b
e
rg
,
M.
A
n
g
e
lo
v
a
,
a
n
d
J.
K
o
h
sta
ll
,
“
Cl
u
ste
r
V
a
li
d
a
ti
o
n
M
e
a
su
re
s
f
o
r
L
a
b
e
l
No
ise
F
il
terin
g
”,
In
ter
n
a
t
io
n
a
l
C
o
n
fer
e
n
c
e
o
n
I
n
tel
li
g
e
n
t
S
y
ste
ms
,
p
p
.
1
0
9
-
1
1
6
,
2
0
1
8
.
[5
]
B.
F
re
n
a
y
a
n
d
M
.
V
e
rley
s
e
n
,
“
Clas
sif
ic
a
ti
o
n
in
th
e
P
re
se
n
c
e
o
f
L
a
b
e
l
No
ise
:
a
S
u
rv
e
y
”
,
IEE
E
T
ra
n
sa
c
ti
o
n
s
o
n
Ne
u
ra
l
Ne
two
rk
s a
n
d
L
e
a
rn
i
n
g
S
y
ste
ms
,
v
o
l.
2
5
,
n
o
.
5
,
p
p
.
8
4
5
-
8
6
9
,
2
0
1
4
.
[6
]
M
.
Ba
sn
e
r,
e
t
a
l.
,
“
A
u
d
it
o
ry
a
n
d
n
o
n
-
a
u
d
i
to
ry
e
ffe
c
ts
o
f
n
o
ise
o
n
h
e
a
lt
h
”,
L
a
n
c
e
t
,
p
p
.
1
3
2
5
–
1
3
3
2
,
2
0
1
4
.
[7
]
M
.
P
a
m
p
a
k
a
,
G
.
Hu
tch
e
so
n
a
n
d
J.
W
il
li
a
m
s
,
“
Ha
n
d
li
n
g
m
issin
g
d
a
ta:
a
n
a
ly
sis
o
f
a
c
h
a
ll
e
n
g
in
g
d
a
ta
se
t
u
sin
g
m
u
lt
ip
le i
m
p
u
tatio
n
”
,
I
n
ter
n
a
ti
o
n
a
l
J
o
u
rn
a
l
o
f
Res
e
a
rc
h
a
n
d
M
e
th
o
d
i
n
E
d
u
c
a
ti
o
n
,
v
o
l.
3
9
,
n
o
.
1
,
p
p
.
19
-
37
,
2
0
1
6
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
5
0
2
-
4752
I
n
d
o
n
esia
n
J
E
lec
E
n
g
&
C
o
m
p
Sci,
Vo
l.
17
,
No
.
1
,
J
an
u
ar
y
20
20
:
5
4
3
-
550
550
[8
]
R.
T
.
O'
Ne
il
l
a
n
d
R.
T
e
m
p
le,
“
T
h
e
P
re
v
e
n
ti
o
n
a
n
d
T
re
a
tme
n
t
o
f
M
issin
g
Da
ta
in
Cli
n
ica
l
Tri
a
ls:
A
n
F
D
A
P
e
rsp
e
c
ti
v
e
o
n
th
e
Im
p
o
rtan
c
e
o
f
De
a
li
n
g
W
it
h
It”,
Cli
n
ica
l
Ph
a
rm
a
c
o
lo
g
y
a
n
d
T
h
e
ra
p
e
u
t
ics
,
v
o
l.
9
1
,
n
o
.
3
,
p
p
.
5
5
0
-
5
5
4
,
2
0
1
2
.
[9
]
J.
D.
Dz
iu
ra
,
L
.
A
.
P
o
st,
Q.
Zh
a
o
,
Z.
F
u
,
a
n
d
P
.
P
e
d
u
z
z
i,
“
S
trate
g
ies
f
o
r
d
e
a
li
n
g
w
it
h
M
issin
g
d
a
ta
in
c
li
n
ica
l
tri
a
ls:
F
ro
m
d
e
sig
n
to
A
n
a
ly
sis
”
,
Y
a
le Jo
u
rn
a
l
o
f
Bi
o
l
o
g
y
a
n
d
M
e
d
icin
e
,
v
o
l.
8
6
,
p
p
.
3
4
3
-
3
5
8
,
2
0
1
3
.
[1
0
]
C.
M
a
ll
in
c
k
ro
d
t
,
e
t
a
l,
“
Re
c
e
n
t
De
v
e
lo
p
m
e
n
ts
in
th
e
P
re
v
e
n
ti
o
n
a
n
d
T
re
a
t
m
e
n
t
o
f
M
issin
g
D
a
t
a
”
,
T
h
e
ra
p
e
u
ti
c
In
n
o
v
a
ti
o
n
a
n
d
Reg
u
l
a
to
ry
S
c
ie
n
c
e
,
v
o
l.
4
8
,
n
o
.
1
,
p
p
.
6
8
-
8
0
,
2
0
1
3
.
[1
1
]
C.
En
d
e
rs,
“
A
p
p
li
e
d
M
issi
n
g
Da
ta A
n
a
l
y
sis”
,
Gu
il
fo
rd
Pre
ss
,
Ne
w Yo
rk
,
2
0
1
0
.
[1
2
]
C.
V
irm
a
n
i,
A
.
P
il
lai
a
n
d
D.
J
u
n
e
ja,
“
Clu
ste
rin
g
i
n
A
g
g
r
e
g
a
t
e
d
Us
e
r
P
ro
f
il
e
s
a
c
ro
ss
M
u
lt
ip
le
S
o
c
ial
Ne
t
w
o
rk
s”
,
In
ter
n
a
t
io
n
a
l
J
o
u
rn
a
l
o
f
E
lec
trica
l
a
n
d
C
o
mp
u
ter
En
g
in
e
e
rin
g
(
IJ
ECE
),
v
o
l.
7
,
n
o
.
6
,
p
p
.
3
6
9
2
-
3
6
9
9
,
2
0
1
7
.
[1
3
]
K.
S
h
i
a
n
d
L
.
L
i,
“
Hig
h
p
e
rf
o
rm
a
n
c
e
g
e
n
e
ti
c
a
lg
o
rit
h
m
b
a
se
d
tex
t
c
lu
ste
rin
g
u
si
n
g
p
a
rts
o
f
sp
e
e
c
h
a
n
d
o
u
t
li
e
r
e
li
m
in
a
ti
o
n
”
,
Ap
p
li
e
d
In
telli
g
e
n
c
e
,
v
o
l.
3
8
,
p
p
.
5
1
1
-
5
1
9
,
2
0
1
3
.
[1
4
]
I
.
M
.
W
a
rtan
a
,
N
.
P
.
A
g
u
stin
i
a
n
d
J
.
G
.
S
in
g
h
,
“
Op
ti
m
a
l
In
teg
r
a
ti
o
n
o
f
th
e
Re
n
e
w
a
b
le
En
e
rg
y
to
t
h
e
G
rid
b
y
Co
n
sid
e
ri
n
g
S
m
a
ll
S
ig
n
a
l
S
tab
il
it
y
Co
n
stra
in
t”,
In
ter
n
a
ti
o
n
a
l
J
o
u
rn
a
l
o
f
El
e
c
trica
l
a
n
d
C
o
mp
u
t
e
r
En
g
in
e
e
rin
g
(
IJ
ECE
),
v
o
l.
7
,
n
o
.
5
,
p
p
.
2
3
2
9
-
2
3
3
7
,
2
0
1
7
.
[1
5
]
H.
M
.
M
a
n
o
j
a
n
d
A
.
N.
Na
n
d
a
k
u
m
a
r,
“
A
No
v
e
l
Op
ti
m
i
z
a
ti
o
n
to
w
a
rd
s
Hig
h
e
r
Re
li
a
b
il
it
y
in
P
re
d
ictiv
e
M
o
d
e
li
n
g
to
w
a
rd
s
Co
d
e
Re
u
sa
b
il
it
y
”
,
In
ter
n
a
ti
o
n
a
l
J
o
u
r
n
a
l
o
f
El
e
c
trica
l
a
nd
Co
mp
u
ter
En
g
i
n
e
e
rin
g
(
IJ
ECE
),
v
o
l.
7
,
n
o
.
5
,
p
p
.
2
8
5
5
-
2
8
6
2
,
2
0
1
7
.
[1
6
]
S
.
S
u
re
sh
,
S
.
L
a
i,
C.
Ch
e
n
a
n
d
T
.
Ce
li
k
,
“
M
u
lt
isp
e
c
tral
S
a
telli
te
Im
a
g
e
De
n
o
isin
g
v
ia
A
d
a
p
ti
v
e
Cu
c
k
o
o
S
e
a
rc
h
-
Ba
se
d
W
ien
e
r
F
il
ter ”,
IEE
E
T
ra
n
sa
c
ti
o
n
s
o
n
Ge
o
sc
ien
c
e
a
n
d
Rem
o
te S
e
n
si
n
g
,
vo
l.
5
6
,
n
o
.
8
,
p
p
.
4
3
3
4
-
4
3
4
5
,
2
0
1
8
.
[1
7
]
A
.
Bi
f
e
t,
R.
Kirk
b
y
,
G
.
Ho
lme
s
a
n
d
B.
P
f
a
h
rin
g
e
r,
“
M
O
A
:
M
a
ss
iv
e
On
li
n
e
A
n
a
l
y
sis”
,
J
o
u
rn
a
l
o
f
M
a
c
h
in
e
L
e
a
r
n
in
g
Res
e
a
rc
h
,
v
o
l.
1
1
,
p
p
.
1
6
0
1
-
1
6
0
4
,
2
0
1
0
.
[1
8
]
P
.
Ch
a
u
d
h
a
ri,
D.
P
.
Ra
n
a
,
R.
G
.
M
e
h
ta,
N.
J.
M
istry
a
n
d
M
.
M
.
R
a
g
h
u
w
a
n
sh
i,
“
Disc
re
ti
z
a
ti
o
n
o
f
T
e
m
p
o
ra
l
Da
ta:
A
S
u
rv
e
y
”
,
In
ter
n
a
ti
o
n
a
l
J
o
u
rn
a
l
o
f
Co
m
p
u
ter
S
c
ien
c
e
a
n
d
In
f
o
rm
a
ti
o
n
S
e
c
u
rity
,
v
o
l.
1
2
,
n
o
.
2
,
p
p
.
6
6
-
6
9
,
2
0
1
4
.
[1
9
]
S.
Kim
,
C.
A
.
S
u
g
a
r
a
n
d
T
.
R
.
Be
li
n
,
“
Ev
a
lu
a
ti
n
g
m
o
d
e
l
b
a
se
d
im
p
u
tatio
n
m
e
th
o
d
s
f
o
r
m
issi
n
g
c
o
v
a
r
iate
s
in
re
g
re
ss
io
n
m
o
d
e
ls
w
it
h
in
tera
c
ti
o
n
s
”,
S
t
a
ti
stics
i
n
M
e
d
icin
e
,
v
o
l
.
34
,
n
o
.
11
,
p
p
.
1
8
7
6
-
1
8
8
8
,
2
0
1
5
.
[2
0
]
M
.
M
.
M
u
k
a
k
a
,
“
A
G
u
id
e
to
Ap
p
r
o
p
riate
Us
e
o
f
Co
rre
latio
n
Co
e
ff
icie
n
t
in
M
e
d
ica
l
Re
se
a
rc
h
”
,
M
a
la
wi
M
e
d
ica
l
J
o
u
rn
a
l
,
v
o
l
.
2
4
,
n
o
.
3
,
p
p
.
6
9
-
7
1
,
2
0
1
2
.
[2
1
]
T
.
Ch
a
i
a
n
d
R.
R.
Dra
x
l
e
r,
“
Ro
o
t
M
e
a
n
S
q
u
a
re
Err
o
r
(RM
S
E)
o
r
M
e
a
n
A
b
so
lu
te
Err
o
r
(M
A
E)?
-
A
r
g
u
m
e
n
ts
a
g
a
in
st
Av
o
id
in
g
RM
S
E
i
n
th
e
L
it
e
ra
tu
re
”
,
Ge
o
sc
ien
ti
fi
c
M
o
d
e
l
De
v
e
lo
p
m
e
n
t
,
v
o
l.
7
,
p
p
.
1
2
4
7
-
1
2
5
0
,
2
0
1
4
.
[2
2
]
H.
Ka
n
g
,
“
T
h
e
p
re
v
e
n
ti
o
n
a
n
d
h
a
n
d
l
in
g
o
f
th
e
m
issin
g
d
a
ta
”
,
Ko
re
a
n
J
o
u
rn
a
l
o
f
A
n
e
sth
e
sio
l
,
v
o
l.
6
4
,
n
o
.
5
,
402
-
4
0
6
,
2
0
1
3
.
[2
3
]
R.
Ja
b
ra
h
,
e
t
a
l.
,
“
Us
in
g
ra
n
k
e
d
a
u
x
il
iar
y
c
o
v
a
riate
a
s
a
m
o
re
e
ff
ici
e
n
t
sa
m
p
li
n
g
d
e
sig
n
f
o
r
AN
COV
A
m
o
d
e
l:
a
n
a
ly
sis
o
f
a
p
sy
c
h
o
lo
g
ica
l
in
ter
v
e
n
ti
o
n
to
b
u
tt
re
ss
re
sili
e
n
c
e
”
,
Co
mm
u
n
ica
ti
o
n
s
f
o
r
S
ta
t
isti
c
a
l
Ap
p
li
c
a
ti
o
n
s
a
n
d
M
e
th
o
d
s
,
v
o
l
.
2
4
,
p
p
.
2
4
1
-
2
5
4
,
2
0
1
7
.
[2
4
]
S
.
A
.
Cu
lp
e
p
p
e
r
a
n
d
H.
A
g
u
i
n
is,
“
Us
in
g
A
n
a
l
y
sis
o
f
Co
v
a
r
ian
c
e
(
A
NCO
V
A
)
W
it
h
F
a
ll
ib
le
Co
v
a
riate
s
”
,
Psy
c
h
o
lo
g
ica
l
M
e
th
o
d
s
,
v
o
l.
1
6
,
n
o
.
2
,
p
p
.
1
6
6
-
1
7
8
,
2
0
1
1
.
[2
5
]
G
.
G
o
rd
o
n
a
n
d
S
.
Qiu
,
“
A
d
iv
id
e
a
n
d
c
o
n
q
u
e
r
a
lg
o
ri
th
m
f
o
r
e
x
p
lo
it
in
g
p
o
li
c
y
f
u
n
c
ti
o
n
m
o
n
o
to
n
icit
y
”
,
Qu
a
n
t
it
a
t
ive
Eco
n
o
mic
s
,
v
o
l.
9
,
issu
e
.
2
,
p
p
.
5
2
1
-
5
4
0
,
2
0
1
8
.
Evaluation Warning : The document was created with Spire.PDF for Python.