I
nte
rna
t
io
na
l J
o
urna
l o
f
E
lect
rica
l a
nd
Co
m
pu
t
er
E
ng
ineering
(
I
J
E
CE
)
Vo
l.
16
,
No
.
5
,
Octo
b
er
20
26
,
p
p
.
2
8
4
7
~
2
8
5
7
I
SS
N:
2088
-
8
7
0
8
,
DOI
: 1
0
.
1
1
5
9
1
/ijece.
v
16
i
5
.
pp
2
8
4
7
-
2
8
5
7
2847
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ij
ec
e.
ia
esco
r
e.
co
m
AI
-
ba
sed ben
chm
a
rking
f
ra
mewo
r
k f
o
r he
terog
eneo
us ho
spita
l
pha
rma
ceutical d
ema
nd f
o
recas
tin
g
Nendi S
un
en
da
r
1
,
K
urs
ehi F
a
lg
ent
i
1
,
F
a
riza
l G
ina
nja
r
2
1
D
e
p
a
r
t
me
n
t
o
f
I
n
f
o
r
mat
i
o
n
S
y
s
t
e
ms
,
F
a
c
u
l
t
y
o
f
C
o
m
p
u
t
e
r
S
c
i
e
n
c
e
,
U
n
i
v
e
r
si
t
a
s
N
u
s
a
M
a
n
d
i
r
i
,
Ja
k
a
r
t
a
,
I
n
d
o
n
e
si
a
2
H
o
sp
i
t
a
l
M
a
n
a
g
e
me
n
t
I
n
f
o
r
ma
t
i
o
n
S
y
st
e
m I
n
st
a
l
l
a
t
i
o
n
,
R
S
U
P
P
e
r
sa
h
a
b
a
t
a
n
,
Ea
s
t
Ja
k
a
r
t
a
,
I
n
d
o
n
e
si
a
Art
icle
I
nfo
AB
S
T
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
Ma
r
9
,
2
0
2
6
R
ev
is
ed
Au
g
8
,
2
0
2
6
Acc
ep
ted
Au
g
1
9
,
2
0
2
6
Ac
c
u
ra
te
p
h
a
rm
a
c
e
u
ti
c
a
l
d
e
m
a
n
d
fo
re
c
a
stin
g
is
c
h
a
ll
e
n
g
i
n
g
b
e
c
a
u
s
e
h
o
s
p
it
a
l
m
e
d
icin
e
c
o
n
s
u
m
p
ti
o
n
e
x
h
ib
it
s
h
e
tero
g
e
n
e
o
u
s
p
a
tt
e
r
n
s,
i
n
c
lu
d
in
g
se
a
so
n
a
l,
v
o
latil
e
,
a
n
d
i
n
term
it
ten
t
d
e
m
a
n
d
.
T
h
is
stu
d
y
p
r
o
p
o
se
s
th
e
re
g
i
m
e
-
a
wa
r
e
fo
re
c
a
stin
g
fra
m
e
wo
r
k
(RAFF
),
a
re
p
ro
d
u
c
i
b
le
b
e
n
c
h
m
a
rk
i
n
g
a
p
p
ro
a
c
h
th
a
t
c
o
m
b
in
e
s
stru
c
tu
ra
l
d
e
m
a
n
d
d
ia
g
n
o
st
ics
,
lea
k
a
g
e
-
c
o
n
tr
o
ll
e
d
wa
l
k
-
fo
rwa
rd
v
a
li
d
a
ti
o
n
,
sta
ti
stica
l
c
o
m
p
a
riso
n
,
a
n
d
re
g
ime
-
stra
ti
fied
e
v
a
lu
a
ti
o
n
.
Ten
y
e
a
rs
o
f
m
o
n
t
h
l
y
d
e
m
a
n
d
re
c
o
rd
s
c
o
v
e
ri
n
g
3
0
0
p
h
a
rm
a
c
e
u
ti
c
a
l
it
e
m
s
we
re
e
v
a
lu
a
ted
u
sin
g
S
e
a
so
n
a
l
Na
iv
e
,
se
a
so
n
a
l
a
u
to
re
g
re
ss
iv
e
in
teg
ra
ted
m
o
v
in
g
a
v
e
ra
g
e
with
e
x
o
g
e
n
o
u
s
fa
c
t
o
rs
(
S
ARIMAX),
a
n
d
lo
n
g
s
h
o
rt
-
term
m
e
m
o
ry
(LS
TM
)
m
o
d
e
ls.
Re
lativ
e
to
S
e
a
so
n
a
l
Na
iv
e
,
S
ARIMAX
re
d
u
c
e
d
m
e
a
n
a
b
so
lu
te
e
rro
r
b
y
1
8
.
2
0
%
,
a
n
d
r
o
o
t
m
e
a
n
sq
u
a
re
d
e
rro
r
b
y
2
0
.
5
8
%
,
wh
i
le
LS
TM
re
d
u
c
e
d
th
e
m
b
y
1
6
.
4
8
%
a
n
d
2
1
.
7
5
%
,
re
sp
e
c
ti
v
e
ly
.
S
ARIMAX
a
c
h
iev
e
d
th
e
lo
we
st
a
g
g
re
g
a
te
m
e
a
n
a
b
so
lu
te
e
rro
r,
w
h
e
re
a
s
LS
TM
a
c
h
iev
e
d
th
e
l
o
we
st
ro
o
t
m
e
a
n
s
q
u
a
re
d
e
rro
r
a
n
d
g
re
a
ter
sta
b
il
it
y
a
s
v
o
latil
it
y
in
c
re
a
se
d
.
Th
e
re
su
lt
s
s
h
o
w
t
h
a
t
fo
re
c
a
stin
g
p
e
rfo
rm
a
n
c
e
d
e
p
e
n
d
s
o
n
d
e
m
a
n
d
stru
c
t
u
re
,
su
p
p
o
rti
n
g
re
g
ime
-
a
wa
re
m
o
d
e
l
se
lec
ti
o
n
fo
r
h
o
s
p
it
a
l
in
v
e
n
to
r
y
a
n
d
p
r
o
c
u
re
m
e
n
t
p
lan
n
i
n
g
.
K
ey
w
o
r
d
s
:
Fo
r
ec
asti
n
g
L
o
n
g
s
h
o
r
t
-
ter
m
m
e
m
o
r
y
Ph
ar
m
ac
eu
tical
d
em
an
d
SAR
I
MA
X
W
alk
-
f
o
r
war
d
v
alid
atio
n
T
h
is i
s
a
n
o
p
e
n
a
c
c
e
ss
a
rticle
u
n
d
e
r th
e
CC B
Y
-
SA
li
c
e
n
se
.
C
o
r
r
e
s
p
o
nd
ing
A
uth
o
r
:
Nen
d
i Su
n
en
d
a
r
Dep
ar
tm
en
t o
f
I
n
f
o
r
m
atio
n
Sy
s
tem
s
,
Facu
lty
o
f
C
o
m
p
u
ter
S
cien
ce
,
Un
iv
er
s
itas
Nu
s
a
Ma
n
d
ir
i
J
ak
ar
ta,
I
n
d
o
n
esia
E
m
ail: 1
4
2
4
0
0
2
7
@
n
u
s
am
an
d
i
r
i.a
c.
id
1.
I
NT
RO
D
UCT
I
O
N
R
eliab
le
f
o
r
ec
asti
n
g
o
f
p
h
ar
m
ac
eu
tical
d
em
an
d
is
im
p
o
r
t
an
t
f
o
r
m
ain
tain
in
g
a
n
ad
e
q
u
ate
h
o
s
p
ital
in
v
en
to
r
y
wh
ile
av
o
id
in
g
b
o
t
h
s
h
o
r
tag
es a
n
d
ex
ce
s
s
iv
e
s
to
ck
.
Ho
wev
er
,
m
ed
icin
e
c
o
n
s
u
m
p
tio
n
is
n
o
t u
n
if
o
r
m
ac
r
o
s
s
p
r
o
d
u
cts
b
ec
au
s
e
in
d
iv
id
u
al
item
s
m
ay
ex
h
ib
it
d
if
f
er
en
t
lev
els
o
f
v
ar
iab
i
lity
,
in
ter
m
itten
cy
,
s
ea
s
o
n
ality
,
an
d
tem
p
o
r
al
d
e
p
en
d
en
ce
.
C
o
n
s
eq
u
e
n
tly
,
a
p
p
ly
in
g
a
s
in
g
le
f
o
r
ec
asti
n
g
ap
p
r
o
ac
h
to
th
e
e
n
tire
p
h
ar
m
ac
eu
tical
p
o
r
tf
o
lio
m
a
y
co
n
ce
al
s
u
b
s
tan
tial
d
if
f
er
en
ce
s
in
m
o
d
el
p
er
f
o
r
m
a
n
c
e
ac
r
o
s
s
in
d
iv
id
u
al
m
ed
icin
es
[
1
]
–
[
4
]
.
Statis
t
ical
an
d
m
ac
h
in
e
-
lear
n
i
n
g
m
o
d
els
h
av
e
b
ee
n
wid
ely
a
p
p
lied
to
tim
e
-
s
er
ies
f
o
r
ec
asti
n
g
[
5
]
,
[
6
]
.
Seaso
n
al
au
to
r
eg
r
ess
iv
e
in
te
g
r
ated
m
o
v
in
g
a
v
er
ag
e
wit
h
ex
o
g
e
n
o
u
s
f
ac
to
r
s
(
SAR
I
MA
X)
an
d
r
elate
d
s
tatis
t
ical
m
o
d
els
r
em
ain
ef
f
ec
tiv
e
f
o
r
a
p
p
r
o
x
im
ately
lin
ea
r
a
n
d
r
ec
u
r
r
in
g
s
ea
s
o
n
al
p
atter
n
s
[
5
]
,
[
7
]
,
[
8
]
,
wh
ile
lo
n
g
s
h
o
r
t
-
ter
m
m
em
o
r
y
(
L
S
T
M)
ca
n
ca
p
tu
r
e
n
o
n
lin
ea
r
t
em
p
o
r
al
d
ep
en
d
en
cies
an
d
c
o
m
p
lex
s
eq
u
en
tial
r
elatio
n
s
h
ip
s
[
9
]
–
[
1
2
]
.
C
o
m
p
a
r
ativ
e
s
tu
d
ies
s
h
o
w
th
at
au
to
r
eg
r
ess
iv
e
in
teg
r
ated
m
o
v
i
n
g
a
v
er
ag
e
(
AR
I
MA
)
-
f
am
ily
an
d
L
STM
-
f
am
ily
m
o
d
els
m
ay
alter
n
ate
in
p
er
f
o
r
m
an
ce
d
ep
e
n
d
in
g
o
n
tim
e
-
s
er
ie
s
ch
ar
ac
ter
is
tics
an
d
ev
alu
atio
n
m
etr
ics
[
7
]
,
[
8
]
,
[
1
3
]
.
Ho
wev
er
,
m
o
d
el
co
m
p
lex
ity
d
o
es
n
o
t
n
e
ce
s
s
ar
ily
g
u
ar
a
n
tee
b
etter
f
o
r
ec
asti
n
g
p
er
f
o
r
m
an
ce
.
R
an
d
o
m
tr
ai
n
-
test
s
p
litt
in
g
m
ay
i
n
tr
o
d
u
ce
tem
p
o
r
al
leak
a
g
e
[
1
4
]
–
[
1
6
]
,
ag
g
r
eg
ate
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8
7
0
8
I
n
t J E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
16
,
No
.
5
,
Octo
b
e
r
20
26
:
2
8
4
7
-
2
8
5
7
2848
m
etr
ics
m
ay
co
n
ce
al
item
-
l
ev
el
h
eter
o
g
en
eity
,
an
d
a
v
e
r
ag
e
-
er
r
o
r
co
m
p
a
r
is
o
n
s
m
ay
o
v
e
r
s
tate
m
o
d
el
s
u
p
er
io
r
ity
[
1
7
]
–
[
2
0
]
.
Pre
v
io
u
s
s
tu
d
ies
h
av
e
em
p
h
asized
co
n
tex
t
-
d
e
p
en
d
e
n
t
m
o
d
el
s
elec
tio
n
,
in
ter
m
itten
t
d
em
an
d
,
tem
p
o
r
al
v
alid
atio
n
,
an
d
o
p
e
r
atio
n
ally
r
ea
lis
tic
ev
alu
atio
n
[
2
]
,
[
1
4
]
,
[
1
5
]
,
[
2
1
]
.
Nev
er
th
eless
,
v
o
latilit
y
,
s
ea
s
o
n
ality
,
an
d
in
ter
m
itten
cy
ar
e
n
o
t
co
n
s
is
ten
tly
in
teg
r
ate
d
in
to
m
o
d
el
b
en
ch
m
a
r
k
in
g
,
wh
ile
p
o
r
tf
o
lio
-
le
v
el
m
ea
s
u
r
es
m
ay
h
i
d
e
d
if
f
er
en
c
es
am
o
n
g
p
h
ar
m
ac
eu
tical
item
s
.
Statis
tical
s
ig
n
if
ican
ce
m
ay
also
b
e
r
ep
o
r
ted
with
o
u
t
co
n
s
id
er
in
g
ef
f
ec
t
m
ag
n
itu
d
e
[
2
2
]
.
T
h
is
cr
ea
tes
a
m
eth
o
d
o
l
o
g
ical
g
ap
f
o
r
h
ete
r
o
g
en
e
o
u
s
h
o
s
p
ital
p
h
ar
m
ac
eu
tical
d
em
an
d
.
T
o
ad
d
r
ess
th
is
g
ap
,
th
is
s
t
u
d
y
p
r
o
p
o
s
es
th
e
r
eg
im
e
-
aw
ar
e
f
o
r
ec
asti
n
g
f
r
am
ewo
r
k
(
R
AFF),
a
r
ep
r
o
d
u
cib
le
b
en
ch
m
ar
k
in
g
f
r
am
ewo
r
k
th
at
i
n
teg
r
ates
s
tr
u
ctu
r
al
d
em
an
d
d
ia
g
n
o
s
tics
,
leak
ag
e
-
c
o
n
tr
o
lle
d
tem
p
o
r
al
v
alid
atio
n
,
s
tatis
tical
co
m
p
ar
is
o
n
,
an
d
r
eg
im
e
-
s
p
ec
if
ic
r
o
b
u
s
tn
ess
an
aly
s
is
.
Ph
ar
m
ac
eu
tical
s
er
ies
ar
e
ch
ar
ac
ter
ize
d
b
y
v
o
latilit
y
,
s
ea
s
o
n
ality
s
tr
en
g
th
,
an
d
i
n
ter
m
itten
cy
,
wh
ile
Seaso
n
al
Naiv
e,
SAR
I
MA
X
,
an
d
L
STM
ar
e
ev
alu
ate
d
u
s
in
g
id
en
tical
ex
p
an
d
in
g
-
win
d
o
w
walk
-
f
o
r
war
d
v
alid
atio
n
.
M
o
d
el
p
er
f
o
r
m
an
c
e
is
ass
es
s
ed
u
s
in
g
m
ea
n
ab
s
o
lu
t
e
er
r
o
r
(
MA
E
)
an
d
r
o
o
t
m
e
an
s
q
u
ar
ed
e
r
r
o
r
(
R
MSE
)
,
co
n
f
id
en
ce
in
ter
v
als,
W
ilco
x
o
n
s
ig
n
ed
-
r
a
n
k
test
s
,
ef
f
ec
t sizes,
an
d
r
eg
im
e
-
s
tr
atif
ie
d
an
aly
s
is
[
2
3
]
.
T
h
e
n
ew
v
alu
e
o
f
R
AFF
is
to
s
h
if
t
f
o
r
ec
asti
n
g
b
e
n
ch
m
ar
k
i
n
g
f
r
o
m
u
n
iv
e
r
s
al
m
o
d
el
r
an
k
in
g
to
war
d
s
tr
u
ctu
r
ally
co
n
d
itio
n
e
d
m
o
d
el
s
elec
tio
n
.
Usi
n
g
ten
y
ea
r
s
o
f
m
o
n
th
ly
d
e
m
an
d
r
ec
o
r
d
s
co
v
er
in
g
3
0
0
p
h
ar
m
ac
eu
tical
item
s
,
th
e
f
r
a
m
ewo
r
k
c
o
n
n
ec
ts
d
em
an
d
s
tr
u
ctu
r
e,
tem
p
o
r
al
v
alid
atio
n
,
s
tatis
tical
ev
id
en
ce
,
an
d
o
p
er
atio
n
al
m
o
d
el
s
elec
tio
n
to
s
u
p
p
o
r
t d
e
f
en
s
ib
le
h
o
s
p
ita
l in
v
en
to
r
y
an
d
p
r
o
cu
r
em
e
n
t f
o
r
ec
asti
n
g
.
2.
M
E
T
H
O
D
2
.
1
.
Study
des
ig
n
a
nd
RAF
F
wo
rk
f
lo
w
T
h
e
s
tu
d
y
f
o
llo
wed
a
r
et
r
o
s
p
e
ctiv
e
tim
e
-
s
er
ies
b
en
ch
m
ar
k
in
g
d
esig
n
,
with
ea
ch
p
h
a
r
m
ac
e
u
tical
item
r
ep
r
esen
ted
b
y
a
m
o
n
t
h
ly
d
e
m
an
d
s
eq
u
en
ce
.
T
h
e
R
AFF
w
o
r
k
f
lo
w
ag
g
r
eg
ate
d
d
aily
d
is
p
en
s
in
g
r
ec
o
r
d
s
in
to
m
o
n
th
ly
d
e
m
an
d
,
ap
p
lied
tr
ain
in
g
-
win
d
o
w
-
o
n
ly
p
r
ep
r
o
c
ess
in
g
,
ca
lcu
lated
s
tr
u
ctu
r
al
d
escr
ip
to
r
s
,
f
itted
Seaso
n
al
Naiv
e,
SAR
I
MA
X,
an
d
L
STM
m
o
d
els,
p
er
f
o
r
m
ed
ex
p
a
n
d
in
g
-
win
d
o
w
walk
-
f
o
r
war
d
v
alid
atio
n
,
co
m
p
ar
ed
p
air
e
d
er
r
o
r
s
u
s
in
g
co
n
f
id
en
ce
in
ter
v
als,
W
ilco
x
o
n
test
s
,
an
d
ef
f
ec
t
s
izes,
an
d
e
v
alu
ated
r
o
b
u
s
tn
ess
b
y
d
em
an
d
r
eg
im
e
b
ef
o
r
e
g
lo
b
al
m
o
d
el
s
elec
tio
n
.
T
h
is
s
eq
u
en
ce
en
s
u
r
es
th
at
n
o
f
u
tu
r
e
i
n
f
o
r
m
atio
n
is
u
s
ed
d
u
r
in
g
p
r
ep
r
o
ce
s
s
in
g
o
r
m
o
d
e
l
f
itti
n
g
.
T
h
e
o
v
e
r
all
p
r
o
ce
s
s
a
n
d
s
eq
u
en
ce
o
f
th
ese
s
tag
es
ar
e
v
is
u
ally
o
u
tlin
ed
in
Fig
u
r
e
1
.
Fig
u
r
e
1
.
W
o
r
k
f
lo
w
o
f
th
e
r
e
g
im
e
-
awa
r
e
f
o
r
ec
asti
n
g
f
r
am
ew
o
r
k
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J E
lec
&
C
o
m
p
E
n
g
I
SS
N:
2088
-
8
7
0
8
AI
-
b
a
s
ed
b
en
ch
m
a
r
kin
g
fr
a
mewo
r
k
fo
r
h
etero
g
en
eo
u
s
h
o
s
p
ita
l p
h
a
r
ma
ce
u
tica
l
…
(
N
en
d
i
S
u
n
en
d
a
r
)
2849
2
.
2
.
Da
t
a
s
o
urce
,
un
it
o
f
a
na
ly
s
is
,
a
nd
qu
a
ntit
a
t
iv
e
pro
f
ile
T
h
e
d
ataset
co
m
p
r
is
es
1
0
y
ea
r
s
o
f
d
is
p
en
s
in
g
r
ec
o
r
d
s
o
b
tain
e
d
f
r
o
m
a
h
o
s
p
ital
p
h
ar
m
ac
y
in
f
o
r
m
atio
n
s
y
s
tem
in
I
n
d
o
n
esia.
Daily
tr
an
s
ac
tio
n
s
wer
e
ag
g
r
eg
ated
to
ca
le
n
d
ar
m
o
n
th
s
.
T
h
e
an
aly
s
is
in
clu
d
ed
3
0
0
p
h
ar
m
ac
eu
tical
item
s
,
with
a
m
ax
im
u
m
o
f
1
2
0
o
b
s
er
v
atio
n
s
p
er
co
m
p
le
te
item
an
d
3
6
,
0
0
0
n
o
m
in
al
item
-
m
o
n
th
o
b
s
er
v
atio
n
s
.
E
ac
h
r
etain
ed
item
co
n
tain
ed
at
least
3
6
co
n
s
ec
u
tiv
e
m
o
n
th
ly
o
b
s
er
v
atio
n
s
,
wh
ile
item
s
wit
h
m
o
r
e
th
an
2
0
%
s
tr
u
ctu
r
all
y
m
is
s
in
g
r
ec
o
r
d
s
wer
e
ex
clu
d
ed
.
T
h
e
s
ea
s
o
n
al
p
er
io
d
was
1
2
m
o
n
th
s
,
th
e
p
r
i
m
ar
y
v
alid
atio
n
h
o
r
izo
n
was
o
n
e
m
o
n
th
a
h
ea
d
,
an
d
v
o
latilit
y
was
class
if
ied
in
to
lo
w,
m
o
d
er
ate,
an
d
h
ig
h
g
r
o
u
p
s
.
2
.
3
.
Da
t
a
prepro
ce
s
s
ing
a
nd
lea
k
a
g
e
prev
ent
io
n
R
aw
d
is
p
en
s
in
g
r
ec
o
r
d
s
wer
e
tr
an
s
f
o
r
m
ed
in
to
a
c
o
n
tin
u
o
u
s
item
-
m
o
n
th
p
an
el.
E
m
p
ty
item
-
m
o
n
th
ce
lls
wer
e
class
if
ied
as
v
er
if
ied
ze
r
o
-
d
em
an
d
m
o
n
th
s
o
r
u
n
v
er
if
ied
s
o
u
r
ce
-
s
y
s
tem
g
ap
s
.
V
er
if
ied
ze
r
o
s
wer
e
r
etain
ed
,
wh
er
ea
s
u
n
v
e
r
if
ied
g
ap
s
wer
e
n
o
t
au
to
m
atica
lly
co
n
v
er
ted
to
ze
r
o
.
Ser
ies
with
f
ewe
r
th
an
3
6
co
n
s
ec
u
tiv
e
m
o
n
th
ly
o
b
s
er
v
at
io
n
s
o
r
m
o
r
e
th
a
n
2
0
%
s
tr
u
ctu
r
al
m
is
s
in
g
n
ess
wer
e
ex
clu
d
e
d
.
E
x
tr
em
e
p
o
s
itiv
e
v
alu
es
wer
e
win
s
o
r
ized
u
s
in
g
an
u
p
p
er
ca
p
esti
m
ated
with
in
ea
ch
tr
ai
n
in
g
win
d
o
w,
wh
ile
v
er
if
ied
ze
r
o
s
an
d
th
e
lo
wer
tail
wer
e
u
n
ch
an
g
ed
.
Scalin
g
p
a
r
am
eter
s
,
lag
g
ed
p
r
ed
icto
r
s
,
an
d
r
o
llin
g
s
tatis
tics
wer
e
co
n
s
tr
u
cted
o
n
ly
f
r
o
m
th
e
h
is
to
r
ical
tr
ai
n
in
g
win
d
o
w
a
n
d
a
p
p
lied
t
o
th
e
n
e
x
t
-
m
o
n
th
in
p
u
t.
Pre
d
i
ctio
n
s
wer
e
in
v
e
r
s
e
tr
an
s
f
o
r
m
ed
b
ef
o
r
e
MA
E
an
d
R
MSE
ca
lcu
latio
n
.
2
.
4
.
Str
uct
ura
l
dema
nd
des
c
ript
o
rs a
nd
re
g
im
e
t
hresh
o
lds
E
ac
h
item
i w
as c
h
ar
ac
ter
ized
b
y
v
o
latilit
y
,
s
ea
s
o
n
ality
s
tr
en
g
th
,
an
d
in
ter
m
itten
cy
,
=
(
,
,
)
,
T
h
ese
d
escr
ip
to
r
s
wer
e
u
s
ed
d
iag
n
o
s
tically
r
ath
er
t
h
an
as
f
u
tu
r
e
p
r
e
d
icto
r
s
.
Vo
latilit
y
was m
ea
s
u
r
ed
u
s
in
g
th
e
co
ef
f
icien
t o
f
v
ar
iatio
n
=
Seaso
n
ality
s
tr
en
g
th
was e
s
ti
m
ated
f
r
o
m
STL
d
ec
o
m
p
o
s
itio
n
,
an
d
in
ter
m
itten
cy
was m
ea
s
u
r
ed
u
s
in
g
th
e
ze
r
o
-
d
em
an
d
r
atio
.
Fo
r
r
o
b
u
s
tn
ess
an
aly
s
is
,
th
e
3
3
r
d
a
n
d
6
6
th
p
er
ce
n
tiles
wer
e
d
iv
id
e
d
in
t
o
lo
w,
m
o
d
e
r
ate,
an
d
h
ig
h
v
o
latilit
y
g
r
o
u
p
s
u
s
in
g
h
is
to
r
ical
d
ata
o
n
ly
.
Seaso
n
ality
an
d
in
ter
m
itten
cy
r
e
m
ain
ed
co
n
tin
u
o
u
s
d
escr
ip
to
r
s
,
wh
ile
th
e
p
r
in
ci
p
al
s
tr
atif
ied
an
aly
s
is
f
o
cu
s
ed
o
n
v
o
latilit
y
to
av
o
id
s
m
all
a
n
d
u
n
s
tab
le
cr
o
s
s
-
co
m
b
in
atio
n
g
r
o
u
p
s
.
2
.
5
.
F
o
re
ca
s
t
ing
mo
dels
a
nd
co
nfig
ura
t
io
n
T
h
r
ee
f
o
r
ec
asti
n
g
m
o
d
els
wer
e
ev
alu
ated
.
Seaso
n
al
Naiv
e
p
r
ed
icts
m
o
n
t
h
(
t+1
)
u
s
in
g
th
e
o
b
s
er
v
ed
d
em
an
d
f
r
o
m
(
t+1
-
1
2
)
.
SAR
I
MA
X
u
s
ed
th
e
f
o
r
m
(
p
,
d
,
q
)
(
P,
D,
Q)
12
,
wi
th
ca
n
d
id
ate
o
r
d
er
s
ev
alu
ated
b
y
g
r
id
s
ea
r
ch
an
d
th
e
co
n
v
e
r
g
ed
s
p
e
cif
icatio
n
with
th
e
lo
west
Ak
aik
e
I
n
f
o
r
m
atio
n
C
r
iter
io
n
r
et
ain
ed
.
T
h
e
L
STM
m
o
d
el
u
s
ed
a
1
2
-
m
o
n
th
in
p
u
t
s
eq
u
en
ce
,
o
n
e
L
STM
lay
er
with
6
4
h
id
d
en
u
n
its
,
an
d
o
n
e
d
e
n
s
e
o
u
tp
u
t
n
eu
r
o
n
.
T
h
e
f
i
n
al
co
n
f
ig
u
r
atio
n
u
s
ed
Ad
am
with
a
lear
n
in
g
r
ate
o
f
0
.
0
0
1
,
5
0
ep
o
c
h
s
,
an
d
a
b
atc
h
s
ize
o
f
3
2
.
On
e
f
ix
ed
ar
ch
itectu
r
e
was a
p
p
lied
ac
r
o
s
s
all
item
s
.
2
.
6
.
Wa
l
k
-
f
o
rwa
rd
v
a
lid
a
t
io
n pro
ce
du
re
E
v
alu
atio
n
u
s
ed
ex
p
an
d
i
n
g
-
w
in
d
o
w
walk
-
f
o
r
war
d
v
alid
atio
n
.
T
h
e
f
ir
s
t
f
o
r
ec
ast
o
r
i
g
in
w
as
cr
ea
ted
o
n
ly
a
f
ter
at
least
3
6
h
is
to
r
ical
m
o
n
th
s
wer
e
av
ailab
le.
At
o
r
ig
in
t,
all
p
r
e
p
r
o
ce
s
s
in
g
a
n
d
m
o
d
el
f
itti
n
g
u
s
e
d
o
b
s
er
v
atio
n
s
1
,
…
,
;
ea
ch
m
o
d
el
g
e
n
er
ated
a
o
n
e
-
m
o
n
th
-
a
h
ea
d
p
r
e
d
ictio
n
f
o
r
t
+1
;
t
h
e
o
b
s
er
v
ed
v
alu
e
f
o
r
t
+1
was
th
en
r
ev
ea
led
an
d
ad
d
e
d
t
o
th
e
tr
ain
in
g
h
is
to
r
y
;
an
d
th
e
o
r
ig
in
ad
v
a
n
ce
d
b
y
o
n
e
m
o
n
t
h
.
I
d
en
tical
o
r
ig
i
n
s
wer
e
u
s
ed
f
o
r
all
m
o
d
els.
T
h
is
p
r
o
ce
d
u
r
e
s
im
u
lates
o
p
er
at
io
n
al
m
o
n
th
l
y
f
o
r
ec
asti
n
g
a
n
d
p
r
ev
e
n
ts
leak
ag
e
f
r
o
m
r
an
d
o
m
tr
ain
–
test
s
p
litt
in
g
.
Mu
lti
-
s
tep
f
o
r
ec
asts
r
ep
o
r
te
d
later
wer
e
g
en
er
ate
d
o
n
ly
a
f
ter
m
o
d
el
s
elec
tio
n
an
d
wer
e
n
o
t u
s
ed
to
ch
o
o
s
e
t
h
e
win
n
in
g
m
o
d
el
2
.
7
.
E
v
a
lua
t
i
o
n
m
et
rics a
nd
s
t
a
t
is
t
ica
l a
na
ly
s
is
Fo
r
item
i
with
H
i
f
o
r
ec
ast o
r
i
g
in
s
,
ac
cu
r
ac
y
was m
ea
s
u
r
ed
u
s
in
g
MA
E
an
d
R
MSE
:
=
1
∑
|
,
ℎ
−
̂
,
ℎ
|
ℎ
=
1
,
=
√
1
ℎ
∑
(
,
ℎ
−
̂
,
ℎ
)
2
ℎ
=
1
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8
7
0
8
I
n
t J E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
16
,
No
.
5
,
Octo
b
e
r
20
26
:
2
8
4
7
-
2
8
5
7
2850
MA
E
r
ep
r
esen
ts
ty
p
ical
ab
s
o
l
u
te
er
r
o
r
,
wh
er
ea
s
R
MSE
p
lace
s
g
r
ea
ter
weig
h
t
o
n
lar
g
e
m
is
s
es
th
at
ca
n
b
e
o
p
er
atio
n
ally
im
p
o
r
ta
n
t
f
o
r
s
to
ck
o
u
t
an
d
o
v
er
s
to
c
k
r
is
k
.
Po
r
tf
o
lio
m
ea
n
s
,
s
tan
d
ar
d
d
ev
iatio
n
s
,
an
d
9
5
%
co
n
f
id
en
ce
in
te
r
v
als
wer
e
ca
lcu
lated
ac
r
o
s
s
th
e
3
0
0
item
-
lev
el
m
etr
ic
v
alu
es.
Pair
ed
d
if
f
er
en
ce
s
wer
e
s
cr
ee
n
ed
u
s
in
g
th
e
Sh
ap
ir
o
–
W
ilk
test
an
d
ev
alu
ated
with
t
h
e
W
ilco
x
o
n
s
ig
n
ed
-
r
an
k
test
wh
en
n
o
r
m
ality
was
n
o
t c
o
n
s
is
ten
tly
s
atis
f
ied
.
2
.
8
.
E
x
perim
ent
a
l
env
ir
o
nm
ent
a
nd
re
pro
du
cibi
lity
T
h
e
p
i
p
elin
e
was
im
p
lem
en
te
d
in
Py
th
o
n
u
s
in
g
Pan
d
as
an
d
Nu
m
Py
f
o
r
d
ata
p
r
o
ce
s
s
in
g
,
Stats
m
o
d
els
f
o
r
SAR
I
MA
X,
T
en
s
o
r
Flo
w/Ker
as
f
o
r
L
STM
,
Scik
it
-
lear
n
f
o
r
s
ca
lin
g
,
SciPy
f
o
r
s
tatis
tical
tes
ts
,
an
d
Ma
tp
lo
tlib
f
o
r
v
is
u
aliza
tio
n
.
P
er
-
o
r
ig
in
f
o
r
ec
asts
,
item
-
lev
el
er
r
o
r
s
,
m
o
d
el
co
n
f
i
g
u
r
atio
n
s
,
r
eg
im
e
d
escr
ip
to
r
s
,
an
d
s
tatis
tical
-
tes
t
o
u
tp
u
ts
wer
e
ex
p
o
r
ted
as
m
ac
h
i
n
e
-
r
ea
d
a
b
l
e
f
iles
.
T
h
e
an
o
n
y
m
ized
r
ep
r
o
d
u
cib
ilit
y
p
ac
k
a
g
e
is
r
ef
er
en
ce
d
in
th
e
Data
Av
ailab
ilit
y
s
ec
tio
n
s
o
th
at
r
ea
d
er
s
ca
n
in
s
p
ec
t
th
e
item
-
lev
el
e
v
id
en
ce
u
n
d
er
ly
in
g
th
e
ag
g
r
e
g
ate
tab
les.
3.
RE
SU
L
T
S AN
D
D
I
SCU
SS
I
O
N
3
.
1
.
E
m
pirica
l
dema
nd
cha
ra
ct
er
is
t
ics
T
h
is
s
u
b
s
ec
tio
n
p
r
esen
ts
v
is
u
al
ev
id
en
ce
o
f
s
tr
u
ctu
r
al
h
eter
o
g
en
eity
in
h
o
s
p
ital
p
h
ar
m
ac
eu
tical
d
em
an
d
.
T
o
illu
s
tr
ate
th
ese
p
r
o
p
er
ties
,
Fig
u
r
e
2
s
u
m
m
a
r
iz
es
th
e
v
is
u
al
p
atter
n
s
a
n
d
f
r
e
q
u
en
cy
d
is
tr
ib
u
tio
n
o
f
th
e
ev
alu
ated
s
er
ies.
R
ep
r
esen
tativ
e
tim
e
s
er
ie
s
in
Fig
u
r
e
2
(
a
)
an
d
th
e
o
v
er
all
d
em
an
d
d
is
tr
ib
u
tio
n
in
Fig
u
r
e
2
(
b
)
s
h
o
w
h
ete
r
o
g
e
n
eo
u
s
d
em
a
n
d
p
atter
n
s
,
r
an
g
in
g
f
r
o
m
s
tab
le
r
ec
u
r
r
in
g
s
er
ies
to
ir
r
e
g
u
lar
f
lu
ctu
atio
n
s
an
d
in
ter
m
itten
t
d
em
an
d
.
T
h
e
r
i
g
h
t
-
s
k
ewe
d
d
is
tr
ib
u
tio
n
i
n
Fig
u
r
e
2
(
b
)
f
u
r
th
e
r
in
d
icate
s
th
at
m
o
s
t
o
b
s
er
v
atio
n
s
f
all
with
in
a
m
o
d
er
ate
d
em
an
d
r
an
g
e,
wh
ile
f
e
wer
item
s
ex
h
ib
it h
ig
h
e
r
d
em
a
n
d
in
ten
s
ity
.
(
a)
(
b
)
Fig
u
r
e
2
.
Ph
ar
m
ac
eu
tical
ch
ar
ac
ter
is
tics
(
a)
r
ep
r
esen
tativ
e
tim
e
s
er
ies an
d
(
b
)
d
em
an
d
d
is
t
r
ib
u
tio
n
h
is
to
g
r
am
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J E
lec
&
C
o
m
p
E
n
g
I
SS
N:
2088
-
8
7
0
8
AI
-
b
a
s
ed
b
en
ch
m
a
r
kin
g
fr
a
mewo
r
k
fo
r
h
etero
g
en
eo
u
s
h
o
s
p
ita
l p
h
a
r
ma
ce
u
tica
l
…
(
N
en
d
i
S
u
n
en
d
a
r
)
2851
STL
d
ec
o
m
p
o
s
itio
n
r
e
v
ea
ls
g
r
ad
u
al
tr
en
d
s
,
m
o
d
er
ate
s
ea
s
o
n
al
f
lu
ctu
atio
n
s
,
an
d
ir
r
e
g
u
l
ar
r
esid
u
al
s
h
o
ck
s
,
in
d
icatin
g
h
ete
r
o
g
e
n
eo
u
s
tem
p
o
r
al
s
tr
u
ct
u
r
es.
T
h
er
ef
o
r
e,
f
o
r
ec
asti
n
g
p
er
f
o
r
m
an
ce
s
h
o
u
ld
b
e
ev
alu
ated
u
s
in
g
m
u
ltip
le
m
o
d
elin
g
ap
p
r
o
ac
h
es
r
ath
e
r
th
an
r
ely
in
g
o
n
a
s
in
g
le
s
tr
u
ct
u
r
al
ass
u
m
p
tio
n
.
T
h
ese
co
m
p
o
n
en
ts
ar
e
illu
s
tr
ated
in
Fig
u
r
e
3
.
Fig
u
r
e
3
.
STL
d
ec
o
m
p
o
s
itio
n
3
.
2
.
M
o
del
ev
a
lua
t
i
o
n f
ra
m
e
wo
rk
T
h
e
h
eter
o
g
en
eo
u
s
s
tr
u
ctu
r
e
o
f
h
o
s
p
ital
p
h
a
r
m
ac
eu
tical
d
e
m
an
d
p
r
o
d
u
ce
s
s
u
b
s
tan
tial
v
a
r
iatio
n
in
f
o
r
ec
asti
n
g
p
er
f
o
r
m
an
ce
ac
r
o
s
s
d
r
u
g
item
s
.
E
r
r
o
r
d
is
tr
ib
u
tio
n
s
wer
e
th
e
r
ef
o
r
e
an
aly
ze
d
t
o
ev
alu
ate
p
r
ed
ictiv
e
s
tab
ilit
y
an
d
d
is
p
er
s
io
n
ac
r
o
s
s
Seaso
n
al
Naiv
e,
SAR
I
MA
X,
an
d
L
STM
.
Fig
u
r
e
4
s
h
o
ws
th
at
th
e
Seaso
n
al
Naiv
e
b
aselin
e
h
as
t
h
e
wid
es
t
MA
E
d
is
p
er
s
io
n
an
d
lo
n
g
est
r
ig
h
t
-
tail
d
is
tr
ib
u
tio
n
,
wh
er
e
as
SAR
I
MA
X
an
d
L
STM
ex
h
ib
it m
o
r
e
co
n
ce
n
tr
a
ted
er
r
o
r
d
is
tr
ib
u
tio
n
s
.
Fig
u
r
e
4
.
Dis
tr
ib
u
tio
n
o
f
MA
E
ac
r
o
s
s
ev
alu
ated
f
o
r
ec
asti
n
g
m
o
d
els
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8
7
0
8
I
n
t J E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
16
,
No
.
5
,
Octo
b
e
r
20
26
:
2
8
4
7
-
2
8
5
7
2852
Pair
ed
er
r
o
r
c
o
m
p
a
r
is
o
n
s
in
Fig
u
r
e
5
s
h
o
w
s
y
s
tem
atic
r
ed
u
ct
io
n
s
f
o
r
SAR
I
MA
X
an
d
L
ST
M
r
elativ
e
to
S
ea
s
o
n
al
Naiv
e
f
o
r
b
o
th
MA
E
in
Fig
u
r
e
5
(
a)
an
d
R
M
SE
in
Fig
u
r
e
5
(
b
)
,
wh
ile
th
e
SAR
I
MA
X
–
L
ST
M
d
if
f
er
en
ce
s
r
em
ai
n
r
elativ
ely
s
m
all
an
d
ce
n
ter
e
d
n
ea
r
ze
r
o
.
Fig
u
r
e
6
f
u
r
th
e
r
s
h
o
ws
th
at
th
e
b
aselin
e
co
m
p
ar
is
o
n
s
f
o
r
Seaso
n
al
Naiv
e
v
er
s
u
s
SA
R
I
MA
X
in
Fig
u
r
e
6
(
a)
a
n
d
S
ea
s
o
n
al
Na
iv
e
v
er
s
u
s
L
STM
in
Fig
u
r
e
6
(
b
)
ar
e
p
r
e
d
o
m
in
a
n
tly
s
h
if
t
ed
ab
o
v
e
ze
r
o
.
I
n
co
n
tr
ast,
th
e
SAR
I
MA
X
v
er
s
u
s
L
STM
d
is
tr
ib
u
tio
n
in
Fig
u
r
e
6
(
c
)
is
ce
n
ter
e
d
clo
s
e
to
z
er
o
,
in
d
icatin
g
th
at
th
eir
r
elativ
e
ad
v
a
n
tag
e
v
a
r
ies ac
r
o
s
s
p
h
ar
m
ac
eu
tical
item
s
.
(
a)
(
b
)
Fig
u
r
e
5
.
Pair
ed
m
o
d
el
d
if
f
er
e
n
ce
s
(
a)
MA
E
d
if
f
e
r
en
ce
s
ac
r
o
s
s
m
o
d
el
p
air
s
an
d
(
b
)
R
MSE
d
if
f
er
en
ce
s
ac
r
o
s
s
m
o
d
el
p
air
s
(
a)
(
b
)
(
c)
Fig
u
r
e
6
.
Pair
ed
MA
E
d
if
f
er
e
n
ce
s
ac
r
o
s
s
m
o
d
els (
a)
Seaso
n
al
Naiv
e
v
er
s
u
s
SAR
I
MA
X,
(
b
)
Seaso
n
al
Naiv
e
v
er
s
u
s
L
STM
,
an
d
(
c
)
SAR
I
MA
X
v
er
s
u
s
L
STM
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J E
lec
&
C
o
m
p
E
n
g
I
SS
N:
2088
-
8
7
0
8
AI
-
b
a
s
ed
b
en
ch
m
a
r
kin
g
fr
a
mewo
r
k
fo
r
h
etero
g
en
eo
u
s
h
o
s
p
ita
l p
h
a
r
ma
ce
u
tica
l
…
(
N
en
d
i
S
u
n
en
d
a
r
)
2853
3
.
3
.
O
v
er
a
ll
m
o
del per
f
o
rma
nce
co
m
pa
riso
n
Ov
er
all
p
r
ed
ictiv
e
p
er
f
o
r
m
an
c
e
ac
r
o
s
s
th
e
3
0
0
p
h
ar
m
ac
eu
tic
al
item
s
is
s
u
m
m
ar
ized
in
T
ab
le
1
.
B
o
th
ad
v
an
ce
d
m
o
d
els
s
u
b
s
tan
tially
im
p
r
o
v
ed
o
n
th
e
Seaso
n
al
Na
iv
e
b
aselin
e.
SAR
I
MA
X
r
ed
u
ce
d
m
ea
n
MA
E
b
y
1
8
.
2
0
%,
wh
er
ea
s
L
STM
r
ed
u
ce
d
it
b
y
1
6
.
4
8
%.
Fo
r
R
MSE
,
SAR
I
MA
X
ac
h
iev
ed
a
2
0
.
5
8
%
r
ed
u
ctio
n
,
an
d
L
STM
ac
h
iev
ed
th
e
lar
g
est r
e
d
u
ctio
n
,
2
1
.
7
5
%.
SAR
I
MA
X
ac
h
iev
ed
th
e
lo
west
m
ea
n
MA
E
,
wh
ile
L
STM
ac
h
iev
ed
th
e
lo
west
m
ea
n
R
MSE
an
d
R
MSE
d
is
p
er
s
io
n
.
T
h
e
MA
E
d
if
f
er
en
ce
b
etwe
en
SAR
I
MA
X
an
d
L
STM
was
0
.
2
9
0
u
n
it
s
(
2
.
0
6
%),
wh
er
ea
s
R
MSE
f
av
o
r
ed
L
STM
b
y
0
.
2
7
0
u
n
its
(
1
.
4
7
%).
T
h
ese
s
m
all
d
if
f
er
e
n
ce
s
r
eq
u
ir
e
p
air
e
d
s
tatis
tical
an
d
r
eg
im
e
-
s
p
ec
if
ic
in
ter
p
r
etatio
n
r
ath
er
t
h
an
a
claim
o
f
u
n
i
v
er
s
al
s
u
p
er
io
r
ity
.
T
ab
le
1
.
Ov
e
r
all
f
o
r
ec
asti
n
g
p
er
f
o
r
m
a
n
ce
M
o
d
e
l
M
A
E
(
m
e
a
n
±
S
D
)
9
5
%
C
I
M
A
E
r
e
d
u
c
t
i
o
n
R
M
S
E
(
me
a
n
±
S
D
)
9
5
%
C
I
R
M
S
E
r
e
d
u
c
t
i
o
n
S
e
a
so
n
a
l
N
a
i
v
e
1
6
.
8
8
±
4
.
8
7
[
1
6
.
3
3
,
1
7
.
4
3
]
--
2
3
.
0
6
±
5
.
8
2
[
2
2
.
4
0
,
2
3
.
7
2
]
--
S
A
R
I
M
A
X
1
3
.
8
1
±
3
.
4
6
[
1
3
.
4
1
,
1
4
.
2
0
]
1
8
.
2
0
%
1
8
.
3
1
8
±
4
.
1
8
[
1
7
.
8
4
,
1
8
.
7
2
]
2
0
.
5
8
%
LSTM
1
4
.
1
0
±
3
.
4
2
[
1
3
.
7
1
,
1
4
.
4
8
]
1
6
.
4
8
%
1
8
.
0
4
±
3
.
6
6
[
1
7
.
6
3
,
1
8
.
4
6
]
2
1
.
7
5
%
3
.
4
.
St
a
t
is
t
ica
l
s
ig
nifica
nce
t
esting
Pair
ed
h
y
p
o
th
esis
test
in
g
was
co
n
d
u
cted
to
ev
alu
ate
m
o
d
el
d
if
f
er
e
n
ce
s
.
No
r
m
ality
o
f
p
air
ed
f
o
r
ec
ast
-
er
r
o
r
d
if
f
er
e
n
ce
s
was
ex
am
in
ed
u
s
in
g
th
e
Sh
ap
i
r
o
–
W
ilk
test
,
wh
ich
in
d
icate
d
d
ev
iatio
n
s
f
r
o
m
n
o
r
m
ality
in
s
ev
er
al
co
m
p
ar
is
o
n
s
,
p
ar
ticu
lar
ly
Seaso
n
al
Naiv
e
–
SAR
I
MA
X
MA
E
.
T
h
er
ef
o
r
e,
p
ar
am
etr
i
c
p
air
ed
t
-
test
s
wer
e
n
o
t
co
n
s
id
er
ed
a
p
p
r
o
p
r
iate.
T
h
e
d
etailed
r
esu
lts
o
f
th
e
Sh
ap
ir
o
–
W
ilk
n
o
r
m
ality
test
ac
r
o
s
s
all
m
o
d
el
co
m
p
ar
is
o
n
s
an
d
m
e
tr
ics ar
e
r
ep
o
r
ted
in
T
ab
le
2
.
T
h
er
ef
o
r
e,
th
e
W
ilco
x
o
n
s
ig
n
ed
-
r
an
k
test
was
ap
p
lied
as
a
n
o
n
-
p
ar
am
etr
ic
alter
n
ativ
e
f
o
r
ev
alu
atin
g
p
air
ed
m
o
d
el
d
if
f
e
r
en
ce
s
.
T
h
e
W
ilco
x
o
n
test
ass
ess
es
wh
eth
er
th
e
m
ed
ia
n
o
f
t
h
e
p
air
ed
d
if
f
er
en
ce
s
s
ig
n
if
ican
tly
d
ev
iates
f
r
o
m
ze
r
o
with
o
u
t
r
eq
u
ir
in
g
n
o
r
m
ality
ass
u
m
p
tio
n
s
.
T
h
e
s
tatis
t
ical
r
e
s
u
lts
ar
e
p
r
esen
ted
in
T
ab
le
3
.
T
h
er
ef
o
r
e,
th
e
W
ilco
x
o
n
s
ig
n
ed
-
r
a
n
k
test
was
u
s
ed
to
e
v
alu
ate
p
air
ed
m
o
d
el
d
if
f
er
e
n
ce
s
.
B
o
th
SAR
I
MA
X
an
d
L
STM
s
h
o
we
d
lar
g
e
im
p
r
o
v
e
m
en
ts
o
v
er
Seaso
n
al
Naiv
e
(
r
=0
.
6
4
3
–
0
.
8
1
2
,
all
p
<0
.
0
0
1
)
.
T
h
e
SAR
I
MA
X
–
L
STM
d
if
f
er
en
ce
was
s
ig
n
if
ican
t
f
o
r
MA
E
(
p
=0
.
0
2
0
)
b
u
t
s
m
all
i
n
m
a
g
n
itu
d
e
(
r
=
0
.
1
3
4
)
,
wh
ile
th
e
R
MSE
d
if
f
er
e
n
ce
was
n
o
t
s
ig
n
if
ican
t
(
p
=0
.
0
7
9
,
r
=
0
.
1
0
1
)
.
O
v
er
all,
b
o
th
ad
v
a
n
ce
d
m
o
d
els
s
u
b
s
tan
tially
o
u
tp
er
f
o
r
m
th
e
s
ea
s
o
n
al
b
aselin
e,
b
u
t t
h
eir
p
o
r
tf
o
lio
-
le
v
el
d
if
f
er
en
ce
s
r
em
ain
s
m
all.
T
ab
le
2
.
Sh
ap
i
r
o
–
wilk
n
o
r
m
ali
ty
test
f
o
r
p
air
e
d
d
if
f
e
r
en
ce
s
C
o
m
p
a
r
i
so
n
M
e
t
r
i
c
n
W
p
-
v
a
l
u
e
S
e
a
so
n
a
l
N
a
i
v
e
v
s
S
A
R
I
M
A
X
M
A
E
3
0
0
0
.
9
8
4
7
2
0
.
0
0
2
8
0
2
S
e
a
so
n
a
l
N
a
i
v
e
v
s
LST
M
M
A
E
3
0
0
0
.
9
9
6
2
1
0
.
6
9
2
4
0
3
S
A
R
I
M
A
X
v
s L
S
T
M
M
A
E
3
0
0
0
.
9
9
3
0
0
0
.
1
7
3
6
8
0
S
e
a
so
n
a
l
N
a
i
v
e
v
s
S
A
R
I
M
A
X
R
M
S
E
3
0
0
0
.
9
9
3
0
9
0
.
1
8
1
5
8
6
S
e
a
so
n
a
l
N
a
i
v
e
v
s
LST
M
R
M
S
E
3
0
0
0
.
9
9
4
6
2
0
.
3
7
4
4
5
6
S
A
R
I
M
A
X
v
s L
S
T
M
R
M
S
E
3
0
0
0
.
9
9
7
4
0
0
.
9
1
7
3
3
6
T
ab
le
3
.
W
ilco
x
o
n
s
ig
n
ed
-
r
an
k
test
s
an
d
ef
f
ec
t sizes
C
o
m
p
a
r
i
so
n
M
e
t
r
i
c
Z
p
-
v
a
l
u
e
Ef
f
e
c
t
si
z
e
r
M
a
g
n
i
t
u
d
e
S
e
a
so
n
a
l
N
a
i
v
e
v
s
S
A
R
I
M
A
X
M
A
E
-
1
2
.
3
9
7
2
.
7
1
×
1
0
⁻
³⁵
0
.
7
1
6
La
r
g
e
S
e
a
so
n
a
l
N
a
i
v
e
v
s
LST
M
M
A
E
-
1
1
.
1
3
1
8
.
8
8
×
1
0
⁻
²⁹
0
.
6
4
3
La
r
g
e
S
A
R
I
M
A
X
v
s L
S
T
M
M
A
E
-
2
.
3
2
6
0
.
0
2
0
0
.
1
3
4
S
mal
l
S
e
a
so
n
a
l
N
a
i
v
e
v
s
S
A
R
I
M
A
X
R
M
S
E
-
1
4
.
0
6
2
6
.
5
5
×
1
0
⁻
⁴⁵
0
.
8
1
2
La
r
g
e
S
e
a
so
n
a
l
N
a
i
v
e
v
s
LST
M
R
M
S
E
-
1
3
.
7
4
9
5
.
1
7
×
1
0
⁻
⁴³
0
.
7
9
4
La
r
g
e
S
A
R
I
M
A
X
v
s L
S
T
M
R
M
S
E
-
1
.
7
5
7
0
.
0
7
9
0
.
1
0
1
S
mal
l
3
.
5
.
I
t
e
m
-
lev
el
cr
o
s
s
-
m
o
del a
na
ly
s
is
An
item
-
lev
el
cr
o
s
s
-
m
o
d
el
an
aly
s
is
was
co
n
d
u
cted
to
d
eter
m
in
e
wh
eth
er
p
o
r
tf
o
lio
a
v
er
a
g
es
co
n
ce
al
s
y
s
tem
atic
d
if
f
er
en
ce
s
am
o
n
g
p
h
ar
m
ac
e
u
tical
s
er
ies.
All
m
o
d
els
wer
e
ev
alu
ated
o
n
th
e
s
am
e
item
s
an
d
f
o
r
ec
ast
o
r
ig
in
s
,
allo
win
g
d
ir
ec
t
co
m
p
ar
is
o
n
u
n
d
er
in
ter
m
itten
t
u
s
ag
e,
s
ea
s
o
n
al
v
ar
iab
ilit
y
,
an
d
v
o
latilit
y
s
h
if
ts
.
T
h
e
r
an
k
-
b
ased
s
u
m
m
a
r
y
in
T
ab
le
4
.
is
alig
n
ed
with
th
e
ag
g
r
e
g
ate
m
etr
ics.
SAR
I
MA
X
r
an
k
s
f
ir
s
t
o
n
MA
E
,
wh
er
ea
s
L
STM
r
a
n
k
s
f
ir
s
t
o
n
R
MSE
.
T
h
eir
m
ea
n
ac
cu
r
ac
y
r
a
n
k
is
th
er
ef
o
r
e
tied
,
an
d
th
e
f
in
al
d
ep
lo
y
m
e
n
t
d
ec
is
io
n
m
u
s
t
in
co
r
p
o
r
ate
er
r
o
r
d
is
p
er
s
io
n
a
n
d
r
eg
im
e
r
o
b
u
s
tn
ess
r
ath
er
th
an
an
in
co
n
s
is
ten
t
claim
th
at
L
STM
r
an
k
s
f
ir
s
t o
n
ev
er
y
m
etr
ic.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8
7
0
8
I
n
t J E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
16
,
No
.
5
,
Octo
b
e
r
20
26
:
2
8
4
7
-
2
8
5
7
2854
T
h
e
r
an
k
-
b
ased
s
u
m
m
ar
y
in
T
ab
le
5
.
s
h
o
ws
th
at
SA
R
I
MA
X
r
an
k
s
f
ir
s
t
o
n
MA
E
,
wh
ile
L
S
T
M
r
an
k
s
f
ir
s
t
o
n
R
MSE
,
r
esu
ltin
g
in
a
tied
m
ea
n
ac
cu
r
ac
y
r
an
k
.
T
h
er
ef
o
r
e,
g
lo
b
al
m
o
d
el
s
el
ec
tio
n
s
h
o
u
ld
also
co
n
s
id
er
er
r
o
r
d
is
p
er
s
io
n
a
n
d
r
eg
im
e
r
o
b
u
s
tn
ess
.
T
h
e
r
an
k
tim
e
d
em
o
n
s
tr
ates
th
at
R
AFF
d
o
es
n
o
t
r
ely
o
n
a
s
in
g
le
m
etr
ic.
L
STM
is
co
n
s
i
d
er
ed
f
o
r
d
ep
lo
y
m
en
t
b
ec
au
s
e
R
MSE
p
lace
s
g
r
ea
ter
weig
h
t
o
n
lar
g
e
er
r
o
r
s
an
d
its
r
elativ
e
p
er
f
o
r
m
an
ce
im
p
r
o
v
es
in
m
o
d
er
ate
-
an
d
h
ig
h
-
v
o
latilit
y
r
eg
im
es,
r
ath
er
th
an
b
ec
au
s
e
it
d
o
m
in
ates
SAR
I
MA
X
o
n
m
ea
n
MA
E
.
T
ab
le
4
.
Me
an
MA
E
an
d
R
M
SE
b
y
v
o
latilit
y
class
V
o
l
a
t
i
l
i
t
y
c
l
a
ss
M
o
d
e
l
M
e
a
n
M
A
E
M
e
a
n
R
M
S
E
Lo
w
V
o
l
a
t
i
l
i
t
y
S
e
a
so
n
a
l
N
a
i
v
e
1
4
.
7
2
1
9
.
8
5
S
A
R
I
M
A
X
1
2
.
1
8
1
6
.
7
4
LSTM
1
2
.
3
5
1
6
.
5
2
M
o
d
e
r
a
t
e
V
o
l
a
t
i
l
i
t
y
S
e
a
so
n
a
l
N
a
i
v
e
1
6
.
9
5
2
3
.
4
7
S
A
R
I
M
A
X
1
3
.
8
4
1
8
.
9
2
LSTM
1
3
.
6
2
1
8
.
4
1
H
i
g
h
V
o
l
a
t
i
l
i
t
y
S
e
a
so
n
a
l
N
a
i
v
e
1
9
.
8
4
2
8
.
3
6
S
A
R
I
M
A
X
1
6
.
5
7
2
3
.
1
4
LSTM
1
5
.
9
1
2
2
.
3
8
T
ab
le
5
.
Ag
g
r
eg
ate
ac
c
u
r
ac
y
r
an
k
in
g
M
o
d
e
l
M
A
E
r
a
n
k
R
M
S
E
r
a
n
k
M
e
a
n
r
a
n
k
R
o
b
u
st
n
e
ss
i
n
t
e
r
p
r
e
t
a
t
i
o
n
S
A
R
I
M
A
X
1
2
1
.
5
B
e
st
me
a
n
M
A
E;
s
t
r
o
n
g
e
s
t
i
n
l
o
w
v
o
l
a
t
i
l
i
t
y
LSTM
2
1
1
.
5
B
e
st
me
a
n
R
M
S
E
;
s
t
r
o
n
g
e
st
st
a
b
i
l
i
t
y
a
s v
o
l
a
t
i
l
i
t
y
r
i
ses
S
e
a
so
n
a
l
N
a
i
v
e
3
3
3
.
0
H
i
g
h
e
s
t
e
r
r
o
r
a
n
d
d
i
s
p
e
r
si
o
n
3
.
6
.
Reg
i
m
e
-
s
t
ra
t
if
ied r
o
bu
s
t
nes
s
a
na
ly
s
is
T
o
ass
ess
wh
eth
er
th
e
s
elec
ted
g
lo
b
al
m
o
d
el
m
ain
tain
s
s
tab
le
p
er
f
o
r
m
an
ce
u
n
d
e
r
s
tr
u
ctu
r
ally
h
eter
o
g
en
e
o
u
s
d
em
an
d
co
n
d
i
tio
n
s
,
a
r
e
g
im
e
-
s
tr
atif
ied
r
o
b
u
s
tn
ess
an
aly
s
is
was
co
n
d
u
ct
ed
u
s
in
g
v
o
latilit
y
-
b
ased
s
eg
m
en
tatio
n
.
Dr
u
g
ite
m
s
wer
e
ca
teg
o
r
ize
d
in
to
lo
w
,
m
o
d
e
r
ate,
an
d
h
i
g
h
v
o
latilit
y
class
es
b
ased
o
n
co
ef
f
icien
t
-
of
-
v
ar
iatio
n
th
r
esh
o
ld
s
.
SAR
I
MA
X
p
er
f
o
r
m
s
s
lig
h
tly
b
etter
in
lo
w
-
v
o
latilit
y
s
ettin
g
s
,
wh
ile
L
STM
r
em
ain
s
co
m
p
etitiv
e
an
d
d
eg
r
ad
es
less
as
v
o
latilit
y
in
cr
ea
s
es.
L
STM
ac
h
iev
es
th
e
lo
west
R
MSE
in
m
o
d
er
ate
-
an
d
h
i
g
h
-
v
o
latilit
y
r
eg
im
es,
in
d
icatin
g
b
etter
co
n
tr
o
l o
f
lar
g
e
er
r
o
r
s
an
d
s
u
p
p
o
r
tin
g
its
s
elec
tio
n
as th
e
g
lo
b
al
f
o
r
ec
asti
n
g
m
o
d
el.
3
.
7
.
G
l
o
ba
l
mo
del selec
t
io
n
Glo
b
al
m
o
d
el
s
elec
tio
n
in
teg
r
ated
ag
g
r
e
g
ate
ac
cu
r
ac
y
,
s
tatis
tical
s
ig
n
if
ican
ce
,
m
o
d
el
r
an
k
in
g
,
an
d
r
eg
im
e
-
s
tr
atif
ied
r
o
b
u
s
tn
ess
.
B
o
th
SAR
I
MA
X
an
d
L
S
T
M
o
u
tp
e
r
f
o
r
m
ed
Seaso
n
al
Naiv
e.
SAR
I
MA
X
ac
h
iev
ed
a
s
lig
h
tly
l
o
wer
m
e
an
MA
E
,
wh
ile
L
STM
s
h
ar
e
d
th
e
b
est
m
ea
n
ac
cu
r
ac
y
r
a
n
k
,
s
h
o
wed
lo
wer
R
MSE
d
is
p
er
s
io
n
,
an
d
r
em
ain
ed
s
tab
le
ac
r
o
s
s
v
o
latilit
y
clas
s
es.
T
h
er
ef
o
r
e,
L
STM
was
s
el
ec
ted
as
th
e
g
lo
b
al
f
o
r
ec
asti
n
g
m
o
d
el.
3
.
8
.
I
nte
rpre
t
a
t
io
n
a
nd
co
nn
ec
t
io
n t
o
prio
r
wo
r
k
R
AFF
s
u
p
p
o
r
ts
r
eg
im
e
-
d
ep
e
n
d
en
t
m
o
d
el
s
elec
tio
n
.
SAR
I
MA
X
p
er
f
o
r
m
s
well
in
lo
w
-
v
o
latilit
y
d
em
an
d
,
wh
ile
L
STM
b
etter
co
n
tr
o
ls
lar
g
e
er
r
o
r
s
as
v
o
latilit
y
in
cr
ea
s
es.
T
h
is
ex
p
lain
s
SAR
I
MA
X’
s
s
m
all
MA
E
ad
v
an
tag
e
an
d
L
STM
’
s
p
r
ef
er
en
ce
in
m
o
d
er
ate
-
an
d
h
ig
h
-
v
o
latilit
y
r
eg
im
es.
T
h
e
f
in
d
in
g
s
s
u
p
p
o
r
t
alig
n
in
g
m
o
d
el
ch
o
ice
with
d
em
an
d
ch
a
r
ac
ter
is
tics
r
ath
er
th
an
ass
u
m
in
g
a
u
n
iv
e
r
s
ally
b
est
m
o
d
el.
R
AFF
ex
ten
d
s
th
is
ap
p
r
o
ac
h
b
y
co
m
b
in
in
g
s
tr
u
ctu
r
al
d
escr
ip
to
r
s
,
leak
ag
e
-
co
n
tr
o
lled
v
alid
atio
n
,
an
d
p
air
e
d
ef
f
ec
t
-
s
ize
an
aly
s
is
f
o
r
au
d
itab
le
p
h
a
r
m
ac
eu
tical
d
em
an
d
f
o
r
ec
asti
n
g
d
ec
is
io
n
s
.
3
.
9
.
P
r
a
ct
ica
l,
s
o
ciet
a
l,
a
nd
po
licy
im
pli
ca
t
io
ns
Fo
r
h
o
s
p
ital
a
d
m
in
is
tr
ato
r
s
,
R
AFF
s
u
p
p
o
r
ts
tier
ed
d
ep
lo
y
m
en
t:
SAR
I
MA
X
is
s
u
itab
le
f
o
r
s
tab
le,
s
ea
s
o
n
al
item
s
wh
en
in
ter
p
r
etab
ilit
y
an
d
lo
wer
ab
s
o
lu
te
er
r
o
r
ar
e
p
r
io
r
ities
,
wh
ile
L
STM
ca
n
s
er
v
e
as a
g
lo
b
al
m
o
d
el
wh
e
n
lar
g
e
er
r
o
r
s
ar
e
m
o
r
e
cr
itical.
Fo
r
ec
asts
s
h
o
u
ld
s
u
p
p
o
r
t
p
r
o
cu
r
em
en
t
d
ec
is
io
n
s
alo
n
g
s
id
e
lead
tim
e,
s
af
ety
s
to
ck
,
cr
iticality
,
s
h
elf
life
,
an
d
s
u
p
p
lier
r
eliab
i
lity
.
At
p
o
licy
lev
el,
r
eg
im
e
l
ab
els
ca
n
im
p
r
o
v
e
ev
alu
atio
n
b
y
r
ep
o
r
tin
g
p
er
f
o
r
m
an
ce
ac
co
r
d
i
n
g
to
v
o
latilit
y
an
d
in
ter
m
itten
cy
r
at
h
er
th
an
p
o
r
tf
o
lio
a
v
er
ag
es
alo
n
e.
R
AFF
ca
n
also
s
u
p
p
o
r
t
tr
ac
ea
b
le
o
r
d
er
i
n
g
w
o
r
k
f
l
o
ws
b
y
lin
k
in
g
f
o
r
e
ca
s
ts
with
s
af
ety
s
to
ck
,
r
eo
r
d
er
p
o
in
ts
,
ap
p
r
o
v
al
p
r
o
ce
s
s
es,
an
d
p
u
r
c
h
ase
-
o
r
d
e
r
m
o
n
ito
r
in
g
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J E
lec
&
C
o
m
p
E
n
g
I
SS
N:
2088
-
8
7
0
8
AI
-
b
a
s
ed
b
en
ch
m
a
r
kin
g
fr
a
mewo
r
k
fo
r
h
etero
g
en
eo
u
s
h
o
s
p
ita
l p
h
a
r
ma
ce
u
tica
l
…
(
N
en
d
i
S
u
n
en
d
a
r
)
2855
3
.
1
0
.
F
o
rwa
rd
deplo
y
m
ent
e
v
a
lua
t
io
n
Fo
llo
win
g
L
STM
s
elec
tio
n
,
t
h
e
m
o
d
el
was
r
etr
ain
ed
o
n
th
e
co
m
p
lete
h
is
to
r
ical
d
ataset
to
g
e
n
er
ate
1
-
an
d
3
-
m
o
n
th
-
ah
ea
d
f
o
r
ec
asts
f
o
r
3
0
0
p
h
ar
m
ac
eu
tical
item
s
.
T
h
is
d
ep
lo
y
m
en
t
-
o
r
ien
ted
ev
alu
atio
n
s
u
p
p
o
r
ts
in
v
en
to
r
y
p
lan
n
in
g
a
n
d
p
r
o
cu
r
em
en
t,
with
p
o
r
tf
o
lio
-
lev
el
f
o
r
ec
ast s
tati
s
tics
s
u
m
m
ar
ized
in
T
ab
le
6
.
T
h
e
1
-
m
o
n
th
-
ah
ea
d
f
o
r
ec
ast
p
r
o
d
u
ce
d
a
m
ea
n
p
r
ed
icted
d
e
m
an
d
o
f
3
6
.
1
1
u
n
its
(
SD
=
8
.
0
4
)
,
w
h
ile
th
e
3
-
m
o
n
th
h
o
r
izo
n
r
em
ain
e
d
co
m
p
ar
ab
le
at
3
6
.
2
7
u
n
its
with
s
lig
h
tly
h
i
g
h
er
d
is
p
er
s
io
n
(
SD
=
8
.
8
7
)
.
T
h
e
s
im
ilar
ity
b
etwe
en
s
h
o
r
t
-
an
d
m
ed
iu
m
-
te
r
m
m
ea
n
f
o
r
ec
a
s
ts
in
d
icate
s
th
at
th
e
m
o
d
el
d
o
es
n
o
t
ex
h
i
b
it
s
y
s
tem
atic
d
r
if
t o
v
er
tim
e.
T
ab
le
6
.
Po
r
tf
o
lio
-
lev
el
f
o
r
ec
a
s
t su
m
m
ar
y
H
o
r
i
z
o
n
M
e
a
n
F
o
r
e
c
a
s
t
M
e
d
i
a
n
F
o
r
e
c
a
s
t
S
t
d
.
D
e
v
i
a
si
M
i
n
M
a
x
1
-
M
o
n
t
h
A
h
e
a
d
36
36
8
.
0
4
5
57
3
-
M
o
n
t
h
A
h
e
a
d
37
37
8
.
8
7
12
63
Dir
ec
tio
n
al
ch
an
g
es
wer
e
n
ea
r
ly
b
alan
ce
d
,
with
1
5
2
item
s
in
cr
ea
s
in
g
(
5
0
.
6
7
%)
an
d
1
4
8
d
ec
r
ea
s
in
g
(
4
9
.
3
3
%),
as
d
etailed
in
T
ab
l
e
7
.
T
h
e
m
ea
n
r
elativ
e
ch
a
n
g
e
was
-
0
.
0
2
%
(
SD
=
4
.
7
3
%).
Fo
r
ec
asts
r
em
ain
ed
with
in
h
is
to
r
ically
p
lau
s
ib
le
r
a
n
g
es,
s
u
p
p
o
r
tin
g
s
tab
le
s
h
o
r
t
-
an
d
m
ed
i
u
m
-
ter
m
i
n
v
en
to
r
y
p
l
an
n
in
g
.
T
ab
le
7
.
Dir
ec
tio
n
al
c
h
an
g
e
b
etwe
en
f
o
r
ec
ast h
o
r
izo
n
s
C
a
t
e
g
o
r
y
C
o
u
n
t
P
r
o
p
o
r
t
i
o
n
I
n
c
r
e
a
se
1
5
2
5
0
.
6
7
%
D
e
c
r
e
a
s
e
1
4
8
4
9
.
3
3
%
N
o
C
h
a
n
g
e
0
0
.
0
0
%
3
.
1
1
.
L
im
it
a
t
io
ns
a
nd
f
uture
re
s
ea
rc
h
Sev
er
al
lim
itatio
n
s
r
em
ain
.
T
h
e
s
tu
d
y
u
s
es
m
o
n
th
ly
d
ata
f
r
o
m
o
n
e
h
o
s
p
ital
an
d
ev
al
u
ates
o
n
ly
t
h
r
ee
m
o
d
els.
T
h
e
p
r
im
ar
y
r
e
g
im
e
a
n
aly
s
is
f
o
cu
s
es
o
n
v
o
latilit
y
,
p
r
ep
r
o
ce
s
s
in
g
s
en
s
itiv
ity
was
n
o
t
f
u
lly
test
ed
,
an
d
f
o
r
ec
ast
ac
cu
r
ac
y
was
n
o
t
d
ir
ec
tly
lin
k
ed
to
in
v
e
n
to
r
y
o
r
p
atien
t
-
s
af
ety
o
u
tc
o
m
es.
Fu
tu
r
e
r
esear
ch
s
h
o
u
ld
ev
alu
ate
R
AFF
o
n
m
u
lti
-
h
o
s
p
ital
an
d
h
ig
h
er
-
f
r
eq
u
en
cy
d
ata,
b
r
o
a
d
er
m
o
d
el
f
am
ilies
,
an
d
p
r
ep
r
o
ce
s
s
in
g
s
en
s
itiv
ity
.
Fu
r
th
er
s
tu
d
ies
s
h
o
u
ld
also
co
n
n
ec
t
f
o
r
ec
asti
n
g
with
in
v
en
to
r
y
o
p
tim
iz
atio
n
,
p
r
o
cu
r
em
en
t
co
n
s
tr
ain
ts
,
an
d
o
p
er
atio
n
al
o
u
tco
m
es
s
u
ch
as
s
to
ck
av
ai
lab
ilit
y
,
waste,
tr
ac
ea
b
ilit
y
,
a
n
d
clin
ical
-
s
er
v
ice
co
n
tin
u
ity
.
4.
CO
NCLU
SI
O
N
T
h
is
s
tu
d
y
in
tr
o
d
u
ce
d
R
AFF,
a
r
eg
im
e
-
awa
r
e
an
d
leak
ag
e
-
co
n
tr
o
lled
b
en
ch
m
ar
k
in
g
f
r
a
m
ewo
r
k
f
o
r
h
eter
o
g
en
e
o
u
s
h
o
s
p
ital
p
h
ar
m
ac
eu
tical
d
em
an
d
.
Acr
o
s
s
3
0
0
item
s
,
SAR
I
MA
X
ac
h
iev
ed
th
e
lo
west
ag
g
r
eg
ate
MA
E
,
wh
ile
L
STM
ac
h
iev
ed
th
e
lo
west
ag
g
r
eg
ate
R
MSE
an
d
g
r
ea
te
r
s
tab
ilit
y
as
v
o
latilit
y
in
cr
ea
s
ed
.
L
STM
is
th
er
ef
o
r
e
s
u
itab
le
as
a
g
l
o
b
al
f
o
r
ec
asti
n
g
m
o
d
el
wh
e
n
co
n
tr
o
llin
g
lar
g
e
e
r
r
o
r
s
is
im
p
o
r
tan
t,
wh
e
r
ea
s
SAR
I
MA
X
r
em
ain
s
ap
p
r
o
p
r
i
ate
f
o
r
s
tab
le
d
em
a
n
d
r
e
q
u
ir
i
n
g
in
ter
p
r
etab
ilit
y
an
d
l
o
w
ty
p
ical
er
r
o
r
s
.
R
AFF
s
u
p
p
o
r
ts
r
e
g
im
e
d
iag
n
o
s
is
,
wa
lk
-
f
o
r
war
d
ev
al
u
atio
n
,
s
tatis
ti
ca
l
co
m
p
ar
is
o
n
,
an
d
o
p
er
atio
n
al
m
o
d
el
s
elec
tio
n
.
Fu
tu
r
e
r
esear
ch
s
h
o
u
ld
ex
ten
d
th
e
b
e
n
ch
m
a
r
k
to
m
u
ltip
le
h
o
s
p
itals
,
ass
ess
p
r
ep
r
o
ce
s
s
in
g
s
en
s
itiv
ity
,
an
d
ev
alu
ate
d
o
wn
s
tr
ea
m
i
n
v
en
to
r
y
,
p
r
o
cu
r
em
e
n
t,
an
d
p
atien
t
-
s
a
f
ety
o
u
tco
m
es.
ACK
NO
WL
E
DG
M
E
N
T
S
T
h
is
wo
r
k
was
co
n
d
u
cted
a
s
p
ar
t
o
f
Nen
d
i
Su
n
en
d
ar
'
s
m
aster
's
th
esis
.
T
h
e
au
th
o
r
s
th
an
k
th
e
ac
ad
em
ic
s
u
p
er
v
is
o
r
s
f
o
r
m
e
th
o
d
o
lo
g
ical
g
u
id
a
n
ce
an
d
R
SUP
Per
s
ah
ab
atan
,
E
ast
J
ak
ar
ta,
f
o
r
p
r
o
v
id
i
n
g
ac
ce
s
s
to
an
o
n
y
m
ized
h
is
to
r
i
ca
l
p
h
ar
m
ac
e
u
tical
d
em
an
d
d
ata
u
n
d
er
in
s
titu
tio
n
al
p
er
m
is
s
io
n
an
d
ap
p
licab
le
co
n
f
id
en
tiality
a
n
d
d
ata
-
g
o
v
e
r
n
an
ce
r
eq
u
ir
em
en
ts
.
F
UNDING
I
NF
O
R
M
A
T
I
O
N
T
h
is
r
esear
ch
r
ec
eiv
ed
n
o
s
p
e
cif
ic
g
r
an
t
f
r
o
m
an
y
f
u
n
d
in
g
a
g
en
cy
in
th
e
p
u
b
lic,
co
m
m
er
ci
al,
o
r
n
o
t
-
f
o
r
-
p
r
o
f
it secto
r
s
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8
7
0
8
I
n
t J E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
16
,
No
.
5
,
Octo
b
e
r
20
26
:
2
8
4
7
-
2
8
5
7
2856
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
.
Na
m
e
o
f
Aut
ho
r
C
M
So
Va
Fo
I
R
D
O
E
Vi
Su
P
Fu
Nen
d
i Su
n
en
d
a
r
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
Ku
r
s
eh
i Falg
en
ti
✓
✓
✓
✓
✓
✓
Far
izal
Gin
an
jar
✓
✓
✓
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
Au
th
o
r
s
s
tate
n
o
co
n
f
lict o
f
in
t
er
est.
I
NF
O
RM
E
D
CO
NS
E
N
T
T
h
is
s
tu
d
y
u
s
ed
an
o
n
y
m
ized
p
h
ar
m
ac
eu
tical
d
em
an
d
r
ec
o
r
d
s
an
d
d
id
n
o
t
in
v
o
lv
e
d
i
r
ec
t
p
ar
ticip
atio
n
o
f
h
u
m
an
s
u
b
jects
o
r
th
e
c
o
llectio
n
o
f
p
er
s
o
n
ally
i
d
en
tifia
b
le
in
f
o
r
m
atio
n
.
T
h
e
r
ef
o
r
e,
in
f
o
r
m
ed
c
o
n
s
en
t
was
n
o
t a
p
p
licab
le
t
o
th
is
s
tu
d
y
.
E
T
H
I
CAL AP
P
RO
V
AL
T
h
is
s
tu
d
y
u
s
ed
a
n
o
n
y
m
ized
p
h
ar
m
ac
eu
tical
d
em
an
d
r
ec
o
r
d
s
o
b
tain
ed
f
r
o
m
th
e
h
o
s
p
ital
p
h
ar
m
ac
y
in
f
o
r
m
atio
n
s
y
s
tem
an
d
d
i
d
n
o
t
in
v
o
l
v
e
d
i
r
ec
t
p
ar
ticip
atio
n
o
f
h
u
m
a
n
s
u
b
jects
o
r
th
e
u
s
e
o
f
p
atien
t
-
lev
el
o
r
p
er
s
o
n
ally
i
d
en
tifia
b
le
in
f
o
r
m
atio
n
.
Data
ac
ce
s
s
was
g
r
an
te
d
b
y
th
e
h
o
s
p
ital
in
s
titu
tio
n
a
n
d
th
e
r
ec
o
r
d
s
wer
e
u
s
ed
s
o
lely
f
o
r
r
esear
ch
p
u
r
p
o
s
es.
Acc
o
r
d
in
g
to
th
e
ap
p
l
icab
le
in
s
titu
tio
n
al
r
eq
u
ir
em
e
n
ts
,
f
o
r
m
al
eth
ical
ap
p
r
o
v
al
in
v
o
l
v
in
g
h
u
m
an
s
u
b
jects wa
s
n
o
t r
eq
u
ir
ed
f
o
r
th
is
s
tu
d
y
.
DATA AV
AI
L
AB
I
L
I
T
Y
T
h
e
d
ata
s
u
p
p
o
r
tin
g
th
e
f
in
d
i
n
g
s
o
f
th
is
s
tu
d
y
ar
e
av
ailab
l
e
f
r
o
m
t
h
e
co
r
r
esp
o
n
d
i
n
g
au
t
h
o
r
u
p
o
n
r
ea
s
o
n
ab
le
r
eq
u
est.
T
h
e
d
ataset
co
n
s
is
ts
o
f
an
o
n
y
m
ized
p
h
ar
m
ac
eu
tical
d
em
an
d
r
ec
o
r
d
s
o
b
tain
ed
f
r
o
m
th
e
h
o
s
p
ital p
h
ar
m
ac
y
in
f
o
r
m
atio
n
s
y
s
tem
an
d
is
s
u
b
ject
to
in
s
titu
tio
n
al
d
ata
-
ac
ce
s
s
r
estrictio
n
s
.
RE
F
E
R
E
NC
E
S
[
1
]
Y
.
W
a
n
g
,
L.
W
a
n
g
,
a
n
d
M
.
C
h
e
n
,
“
A
mac
h
i
n
e
l
e
a
r
n
i
n
g
f
r
a
m
e
w
o
r
k
f
o
r
p
h
a
r
mac
e
u
t
i
c
a
l
d
e
ma
n
d
f
o
r
e
c
a
s
t
i
n
g
,
”
I
E
EE
A
c
c
e
ss
,
v
o
l
.
8
,
p
p
.
1
6
7
5
6
6
–
1
6
7
5
7
8
,
2
0
2
0
,
d
o
i
:
1
0
.
1
1
0
9
/
A
C
C
ESS
.
2
0
2
0
.
3
0
2
1
9
2
2
.
[
2
]
J.
B
o
y
l
a
n
a
n
d
A
.
S
y
n
t
e
t
o
s
,
I
n
t
e
rm
i
t
t
e
n
t
d
e
m
a
n
d
f
o
r
e
c
a
st
i
n
g
.
W
i
l
e
y
,
2
0
2
1
.
[
3
]
J.
R
.
d
o
R
e
g
o
a
n
d
M
.
A
.
d
e
M
e
s
q
u
i
t
a
,
“
D
e
ma
n
d
f
o
r
e
c
a
s
t
i
n
g
a
n
d
i
n
v
e
n
t
o
r
y
c
o
n
t
r
o
l
:
A
s
i
mu
l
a
t
i
o
n
st
u
d
y
o
n
a
u
t
o
mo
t
i
v
e
sp
a
r
e
p
a
r
t
s
,
”
I
n
t
e
r
n
a
t
i
o
n
a
l
J
o
u
r
n
a
l
o
f
Pr
o
d
u
c
t
i
o
n
E
c
o
n
o
m
i
c
s
,
v
o
l
.
1
6
1
,
p
p
.
1
–
1
6
,
M
a
r
.
2
0
1
5
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
i
j
p
e
.
2
0
1
4
.
1
1
.
0
0
9
.
[
4
]
K
.
S
w
a
m
i
n
a
t
h
a
n
a
n
d
R
.
V
e
n
k
i
t
a
s
u
b
r
a
mo
n
y
,
“
D
e
m
a
n
d
f
o
r
e
c
a
s
t
i
n
g
f
o
r
f
a
sh
i
o
n
p
r
o
d
u
c
t
s:
A
sy
s
t
e
m
a
t
i
c
r
e
v
i
e
w
,
”
I
n
t
e
rn
a
t
i
o
n
a
l
J
o
u
rn
a
l
o
f
Fo
r
e
c
a
st
i
n
g
,
v
o
l
.
4
0
,
n
o
.
1
,
p
p
.
2
4
7
–
2
6
7
,
Ja
n
.
2
0
2
4
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
i
j
f
o
r
e
c
a
st
.
2
0
2
3
.
0
2
.
0
0
5
.
[
5
]
R
.
J
.
H
y
n
d
ma
n
a
n
d
G
.
A
t
h
a
n
a
s
o
p
o
u
l
o
s,
F
o
re
c
a
s
t
i
n
g
:
p
r
i
n
c
i
p
l
e
s
a
n
d
p
r
a
c
t
i
c
e
,
3
r
d
e
d
.
M
e
l
b
o
u
r
n
e
,
A
u
st
r
a
l
i
a
:
O
Te
x
t
s,
2
0
2
1
.
[
6
]
S
.
M
a
k
r
i
d
a
k
i
s
a
n
d
E.
S
p
i
l
i
o
t
i
s,
“
S
t
a
t
i
st
i
c
a
l
a
n
d
ma
c
h
i
n
e
l
e
a
r
n
i
n
g
f
o
r
e
c
a
s
t
i
n
g
m
e
t
h
o
d
s
:
C
o
n
c
e
r
n
s
a
n
d
w
a
y
s
f
o
r
w
a
r
d
,
”
PLo
S
O
N
E
,
v
o
l
.
1
5
,
n
o
.
3
,
p
.
e
0
2
3
0
0
5
6
,
2
0
2
0
,
d
o
i
:
1
0
.
1
3
7
1
/
j
o
u
r
n
a
l
.
p
o
n
e
.
0
2
3
0
0
5
6
.
[
7
]
M
.
B
i
l
g
i
l
i
a
n
d
E.
P
i
n
a
r
,
“
G
r
o
s
s
e
l
e
c
t
r
i
c
i
t
y
c
o
n
su
m
p
t
i
o
n
f
o
r
e
c
a
st
i
n
g
u
si
n
g
LST
M
a
n
d
S
A
R
I
M
A
a
p
p
r
o
a
c
h
e
s:
A
c
a
se
s
t
u
d
y
o
f
Tü
r
k
i
y
e
,
”
E
n
e
r
g
y
,
v
o
l
.
2
8
4
,
p
.
1
2
8
5
7
5
,
D
e
c
.
2
0
2
3
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
e
n
e
r
g
y
.
2
0
2
3
.
1
2
8
5
7
5
.
[
8
]
H
.
Jd
i
a
n
d
N
.
F
a
l
i
h
,
“
C
o
m
p
a
r
i
so
n
o
f
t
i
m
e
s
e
r
i
e
s
t
e
mp
e
r
a
t
u
r
e
p
r
e
d
i
c
t
i
o
n
w
i
t
h
a
u
t
o
-
r
e
g
r
e
ss
i
v
e
i
n
t
e
g
r
a
t
e
d
m
o
v
i
n
g
a
v
e
r
a
g
e
a
n
d
r
e
c
u
r
r
e
n
t
n
e
u
r
a
l
n
e
t
w
o
r
k
,
”
I
n
t
e
rn
a
t
i
o
n
a
l
J
o
u
r
n
a
l
o
f
El
e
c
t
ri
c
a
l
a
n
d
C
o
m
p
u
t
e
r
E
n
g
i
n
e
e
r
i
n
g
,
v
o
l
.
1
4
,
n
o
.
2
,
p
p
.
1
7
7
0
–
1
7
7
8
,
2
0
2
4
,
d
o
i
:
1
0
.
1
1
5
9
1
/
i
j
e
c
e
.
v
1
4
i
2
.
p
p
1
7
7
0
-
1
7
7
8
.
[
9
]
B
.
Li
m
a
n
d
S
.
Zo
h
r
e
n
,
“
Ti
m
e
-
seri
e
s
f
o
r
e
c
a
st
i
n
g
w
i
t
h
d
e
e
p
l
e
a
r
n
i
n
g
:
A
su
r
v
e
y
,
”
Ph
i
l
o
s
o
p
h
i
c
a
l
T
ra
n
s
a
c
t
i
o
n
s
o
f
t
h
e
R
o
y
a
l
S
o
c
i
e
t
y
A:
M
a
t
h
e
m
a
t
i
c
a
l
,
P
h
y
s
i
c
a
l
a
n
d
E
n
g
i
n
e
e
ri
n
g
S
c
i
e
n
c
e
s
,
v
o
l
.
3
7
9
,
n
o
.
2
1
9
4
,
p
.
2
0
2
0
0
2
0
9
,
2
0
2
1
,
d
o
i
:
1
0
.
1
0
9
8
/
r
st
a
.
2
0
2
0
.
0
2
0
9
.
[
1
0
]
H
.
H
e
w
a
m
a
l
a
g
e
,
C
.
B
e
r
g
me
i
r
,
a
n
d
K
.
B
a
n
d
a
r
a
,
“
R
e
c
u
r
r
e
n
t
n
e
u
r
a
l
n
e
t
w
o
r
k
s
f
o
r
t
i
me
seri
e
s
f
o
r
e
c
a
s
t
i
n
g
:
c
u
r
r
e
n
t
st
a
t
u
s
a
n
d
f
u
t
u
r
e
d
i
r
e
c
t
i
o
n
s,
”
I
n
t
e
rn
a
t
i
o
n
a
l
J
o
u
r
n
a
l
o
f
Fo
re
c
a
st
i
n
g
,
v
o
l
.
3
7
,
n
o
.
1
,
p
p
.
3
8
8
–
4
2
7
,
2
0
2
1
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
i
j
f
o
r
e
c
a
st
.
2
0
2
0
.
0
6
.
0
0
8
.
[
1
1
]
S
.
B
e
n
i
d
i
s
a
n
d
O
t
h
e
r
s,
“
N
e
u
r
a
l
f
o
r
e
c
a
st
i
n
g
:
I
n
t
r
o
d
u
c
t
i
o
n
a
n
d
l
i
t
e
r
a
t
u
r
e
o
v
e
r
v
i
e
w
,
”
I
EEE
T
ra
n
s
a
c
t
i
o
n
s
o
n
N
e
u
r
a
l
N
e
t
w
o
r
k
s
a
n
d
L
e
a
r
n
i
n
g
S
y
st
e
m
s
,
v
o
l
.
3
4
,
n
o
.
4
,
p
p
.
1
8
1
0
–
1
8
2
3
,
2
0
2
3
,
d
o
i
:
1
0
.
1
1
0
9
/
TN
N
LS.2
0
2
1
.
3
0
8
0
2
7
2
.
[
1
2
]
K
.
B
a
n
d
a
r
a
,
C
.
B
e
r
g
m
e
i
r
,
a
n
d
H
.
H
e
w
a
mal
a
g
e
,
“
D
e
e
p
n
e
u
r
a
l
n
e
t
w
o
r
k
s
f
o
r
t
i
me
seri
e
s
f
o
r
e
c
a
s
t
i
n
g
:
A
s
u
r
v
e
y
,
”
AC
M
C
o
m
p
u
t
i
n
g
S
u
r
v
e
y
s
,
v
o
l
.
5
5
,
n
o
.
6
,
p
p
.
1
–
3
6
,
2
0
2
2
,
d
o
i
:
1
0
.
1
1
4
5
/
3
5
3
3
3
8
2
.
Evaluation Warning : The document was created with Spire.PDF for Python.