I
nte
rna
t
io
na
l J
o
urna
l o
f
Adv
a
nces in Applie
d Science
s
(
I
J
AAS)
Vo
l.
1
5
,
No
.
2
,
J
u
n
e
20
2
6
,
p
p
.
501
~
5
1
7
I
SS
N:
2252
-
8814
,
DOI
:
1
0
.
1
1
5
9
1
/i
j
aa
s
.
v
1
5
.
i
2
.
pp
501
-
5
1
7
501
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ij
a
a
s
.
ia
esco
r
e.
co
m
Es
ti
m
a
ti
ng
roa
d t
ra
nspo
rt
e
m
iss
io
n
s in Tha
ila
nd usin
g
da
ta
-
driv
en
m
o
de
ling
appro
a
ches
P
ha
cha
y
a
Cha
iw
cha
n
1
,
K
un
y
a
nu
t
h K
ula
rbph
et
t
o
ng
2
1
D
i
g
i
t
a
l
I
n
n
o
v
a
t
i
o
n
M
a
n
a
g
e
me
n
t
P
r
o
g
r
a
m,
F
a
c
u
l
t
y
o
f
S
c
i
e
n
c
e
a
n
d
T
e
c
h
n
o
l
o
g
y
,
S
u
a
n
S
u
n
a
n
d
h
a
R
a
j
a
b
h
a
t
U
n
i
v
e
r
si
t
y
,
B
a
n
g
k
o
k
,
T
h
a
i
l
a
n
d
2
C
o
mp
u
t
e
r
S
c
i
e
n
c
e
P
r
o
g
r
a
m,
F
a
c
u
l
t
y
o
f
S
c
i
e
n
c
e
a
n
d
T
e
c
h
n
o
l
o
g
y
,
S
u
a
n
S
u
n
a
n
d
h
a
R
a
j
a
b
h
a
t
U
n
i
v
e
r
si
t
y
,
B
a
n
g
k
o
k
,
T
h
a
i
l
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
1
0
,
2
0
2
5
R
ev
i
s
ed
Ma
r
1
3
,
2
0
2
6
A
cc
ep
ted
A
p
r
2
2
,
2
0
2
6
T
ra
ff
ic
e
m
issio
n
s
re
m
a
in
a
p
iv
o
tal
f
a
c
to
r
in
Ba
n
g
k
o
k
’
s
a
ir
q
u
a
li
ty
issu
e
s,
p
a
rti
c
u
larly
a
s
th
e
c
it
y
e
x
p
a
n
d
s
a
n
d
v
e
h
icle
ty
p
e
s
b
e
c
o
m
e
m
o
re
v
a
ried
a
n
d
trad
it
io
n
a
l
e
m
issio
n
-
in
v
e
n
to
ry
to
o
ls,
s
u
c
h
a
s
Co
m
p
u
ter
P
ro
g
r
a
m
m
e
to
Ca
lcu
late
Em
issio
n
s
f
ro
m
Ro
a
d
Tran
sp
o
rt
(
CO
P
ERT
)
a
n
d
lab
o
ra
to
r
y
d
riv
in
g
c
y
c
les
,
a
re
u
se
f
u
l
b
u
t
o
f
ten
f
a
ll
sh
o
rt
w
h
e
n
a
p
p
li
e
d
t
o
c
o
n
g
e
ste
d
r
e
a
l
-
w
o
rld
c
o
n
d
i
ti
o
n
s.
T
h
is
w
o
rk
re
sp
o
n
d
s
to
t
h
a
t
g
a
p
b
y
c
o
m
b
in
in
g
lab
o
ra
to
ry
m
e
a
su
re
m
e
n
ts,
re
g
u
lato
ry
e
m
i
ss
io
n
f
a
c
to
rs,
a
n
d
v
e
h
icle
-
re
g
istratio
n
in
f
o
rm
a
ti
o
n
to
e
stim
a
te
CO
2
e
m
is
sio
n
s
a
t
t
h
e
v
e
h
icle
lev
e
l.
Da
ta
f
ro
m
1
,
8
4
6
,
1
1
1
re
g
istere
d
v
e
h
icle
s
(Ja
n
u
a
ry
2
0
2
3
–
Ja
n
u
a
ry
2
0
2
5
)
w
e
re
c
o
n
so
li
d
a
ted
f
ro
m
En
v
iro
n
m
e
n
tal
P
ro
tec
ti
o
n
A
g
e
n
c
y
(
EP
A
)
f
u
e
l
-
e
c
o
n
o
m
y
tes
ts,
COP
ERT
/E
u
r
o
sta
n
d
a
rd
s
,
a
n
d
t
h
e
T
h
a
i
DLT
re
g
istr
y
.
Af
te
r
h
a
rm
o
n
izin
g
a
tt
rib
u
tes
a
c
ro
ss
a
ll
so
u
rc
e
s,
a
d
d
it
i
o
n
a
l
f
e
a
tu
re
s
w
e
r
e
c
re
a
ted
a
n
d
sc
re
e
n
e
d
u
sin
g
re
c
u
rsiv
e
f
e
a
t
u
re
e
li
m
in
a
ti
o
n
(RF
E).
T
h
re
e
m
o
d
e
ls
w
e
r
e
train
e
d
a
n
d
e
v
a
lu
a
ted
u
sin
g
c
o
e
ff
icie
n
t
o
f
d
e
ter
m
in
a
ti
o
n
(R
2
)
,
r
o
o
t
m
e
a
n
sq
u
a
re
e
rro
r
(
RM
S
E
)
,
a
n
d
m
e
a
n
a
b
so
l
u
te
e
rr
o
r
(
M
A
E
)
.
Ca
tBo
o
st
p
r
o
v
id
e
d
th
e
m
o
st
re
li
a
b
le
p
re
d
ictio
n
s
(R
M
S
E
=
2
3
.
4
g
/k
m
;
M
A
E
=
1
4
.
6
g
/k
m
;
R
2
=
0
.
9
2
1
),
f
o
ll
o
w
e
d
b
y
d
e
e
p
n
e
u
ra
l
n
e
tw
o
rk
(DN
N)
(R
2
=
0
.
8
8
7
)
a
n
d
su
p
p
o
rt
v
e
c
to
r
re
g
r
e
ss
io
n
(S
V
R)
(R
2
=
0
.
8
4
6
)
.
RF
E
c
o
n
siste
n
tl
y
id
e
n
ti
f
ied
f
u
e
l
-
c
o
n
su
m
p
ti
o
n
m
e
a
su
re
s,
e
m
is
sio
n
f
a
c
to
rs
,
a
n
d
tailp
ip
e
CO
2
a
s
th
e
m
o
st
in
f
lu
e
n
ti
a
l
v
a
riab
les
a
n
d
t
h
e
re
su
l
ts
p
ro
v
i
d
e
a
c
lea
re
r
p
ictu
re
o
f
h
o
w
v
e
h
icle
tec
h
n
o
l
o
g
ies
c
o
n
tri
b
u
te
to
Ba
n
g
k
o
k
’s
CO
2
b
u
rd
e
n
a
n
d
su
p
p
o
rt
p
o
li
c
y
m
e
a
su
re
s
su
c
h
a
s
e
m
issio
n
tax
e
s,
in
c
e
n
ti
v
e
s
f
o
r
l
o
w
-
e
m
issio
n
v
e
h
icle
s,
a
n
d
m
o
n
i
to
rin
g
o
f
h
ig
h
-
e
m
issio
n
g
ro
u
p
s.
K
ey
w
o
r
d
s
:
Data
-
d
r
iv
e
n
m
o
d
eli
n
g
E
m
is
s
io
n
e
s
ti
m
atio
n
Ma
ch
i
n
e
lear
n
i
n
g
Ur
b
an
p
o
llu
tio
n
Veh
icle
C
O
2
e
m
is
s
io
n
s
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
BY
-
SA
li
c
e
n
se
.
C
o
r
r
e
s
p
o
nd
ing
A
uth
o
r
:
Ku
n
y
a
n
u
th
K
u
lar
b
p
h
etto
n
g
C
o
m
p
u
ter
Scien
ce
P
r
o
g
r
a
m
,
F
ac
u
lt
y
o
f
Scie
n
ce
an
d
T
ec
h
n
o
l
o
g
y
,
S
u
an
Su
n
a
n
d
h
a
R
aj
ab
h
at
Un
i
v
er
s
it
y
1
U
-
T
h
o
n
g
No
k
R
o
ad
,
B
an
g
k
o
k
1
0
3
0
0
,
T
h
ailan
d
E
m
ail:
k
u
n
y
an
u
t
h
.
k
u
@
s
s
r
u
.
ac
.
th
1.
I
NT
RO
D
UCT
I
O
N
T
h
e
tr
an
s
p
o
r
tatio
n
s
ec
to
r
r
em
ain
s
o
n
e
o
f
th
e
m
ai
n
s
o
u
r
ce
s
o
f
g
r
ee
n
h
o
u
s
e
g
a
s
e
m
i
s
s
io
n
s
i
n
T
h
ailan
d
,
an
d
C
O
2
f
r
o
m
v
eh
icle
ex
h
a
u
s
t
co
n
ti
n
u
es
to
d
r
iv
e
n
atio
n
a
l
w
ar
m
i
n
g
tr
en
d
s
at
a
n
o
tice
ab
le
s
ca
le
[
1
]
–
[
4
]
.
A
ir
p
o
ll
u
tio
n
lin
k
ed
to
tr
a
f
f
ic
also
a
f
f
ec
t
s
p
u
b
lic
h
ea
l
th
d
ir
ec
tl
y
.
T
h
e
i
m
p
ac
t
s
ar
e
p
ar
ticu
lar
l
y
s
er
io
u
s
f
o
r
g
r
o
u
p
s
alr
ea
d
y
at
r
is
k
,
in
cl
u
d
in
g
c
h
ild
r
en
,
o
ld
er
ad
u
lts
,
a
n
d
p
eo
p
le
w
it
h
c
h
r
o
n
ic
r
esp
ir
ato
r
y
o
r
ca
r
d
io
v
ascu
lar
ill
n
es
s
es
.
W
o
r
ld
Hea
lth
Or
g
an
izatio
n
(
W
HO)
r
ep
o
r
ts
th
at
o
u
td
o
o
r
air
p
o
llu
tio
n
ac
co
u
n
ted
f
o
r
r
o
u
g
h
l
y
6
8
%
o
f
p
r
em
at
u
r
e
d
ea
th
s
i
n
2
0
1
9
,
m
ai
n
l
y
f
r
o
m
is
c
h
e
m
ic
h
ea
r
t
d
is
ea
s
e,
s
tr
o
k
e,
ch
r
o
n
ic
o
b
s
tr
u
cti
v
e
p
u
l
m
o
n
ar
y
d
i
s
ea
s
e
(
C
OP
D
)
,
lo
w
er
r
esp
ir
ato
r
y
i
n
f
ec
tio
n
s
,
an
d
l
u
n
g
ca
n
ce
r
[
5
]
.
As
o
f
Sep
te
m
b
er
2
0
2
3
,
T
h
ail
an
d
h
ad
m
o
r
e
t
h
a
n
4
4
m
i
llio
n
r
e
g
is
ter
ed
v
e
h
icles
.
W
h
en
co
m
b
i
n
ed
w
it
h
p
er
s
i
s
ten
t
co
n
g
e
s
tio
n
,
s
to
p
-
a
n
d
-
g
o
d
r
iv
in
g
,
an
d
m
i
x
ed
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8814
I
n
t J
A
d
v
A
p
p
l Sci
,
Vo
l.
1
5
,
No
.
2
,
J
u
n
e
2
0
2
6
:
5
0
1
-
517
502
v
eh
ic
le
tech
n
o
lo
g
ies,
th
e
s
e
co
n
d
itio
n
s
m
a
k
e
C
O
2
e
m
is
s
io
n
s
in
B
an
g
k
o
k
f
ar
m
o
r
e
v
ar
iab
le
th
an
n
atio
n
al
av
er
ag
es
w
o
u
ld
s
u
g
g
es
t
[
6
]
.
Su
c
h
cir
cu
m
s
ta
n
ce
s
p
o
in
t
to
th
e
n
ee
d
f
o
r
m
o
r
e
d
etailed
an
d
d
ata
-
d
r
iv
en
e
m
is
s
io
n
-
esti
m
atio
n
ap
p
r
o
ac
h
es
to
s
u
p
p
o
r
t
u
r
b
an
p
lan
n
i
n
g
an
d
ev
id
e
n
ce
-
b
ased
p
o
licy
.
T
h
e
r
ec
en
t
ad
o
p
tio
n
o
f
E
u
r
o
5
an
d
th
e
tr
an
s
itio
n
to
w
ar
d
E
u
r
o
6
s
tan
d
ar
d
s
r
ef
lect
T
h
aila
n
d
’
s
ef
f
o
r
ts
to
r
e
d
u
ce
p
o
llu
ta
n
ts
s
u
c
h
as
NO
x
,
PM
2.
5
,
h
y
d
r
o
ca
r
b
o
n
s
,
an
d
C
O
.
Ho
w
e
v
er
,
th
ese
r
eg
u
latio
n
s
p
r
o
v
id
e
r
elativ
el
y
li
m
ited
g
u
id
an
ce
o
n
r
ea
l
-
w
o
r
ld
CO
2
e
m
is
s
io
n
s
[
7
]
.
P
r
ev
io
u
s
r
e
s
e
ar
c
h
i
n
T
h
a
il
a
n
d
h
a
s
m
a
i
n
l
y
r
el
ied
o
n
d
e
te
r
m
i
n
is
t
ic
to
o
l
s
,
p
ar
ti
c
u
l
ar
l
y
C
o
m
p
u
te
r
P
r
o
g
r
a
m
m
e
to
C
al
c
u
lat
e
E
m
i
s
s
io
n
s
f
r
o
m
R
o
ad
T
r
a
n
s
p
o
r
t
(
C
OP
E
R
T
)
a
n
d
E
u
r
o
4
–
6
e
m
i
s
s
io
n
f
ac
to
r
s
,
wh
i
c
h
e
s
ti
m
a
t
e
e
m
i
s
s
io
n
s
b
a
s
ed
o
n
s
ta
n
d
ar
d
iz
ed
d
r
i
v
i
n
g
c
y
cl
e
s
a
n
d
t
y
p
ic
al
f
u
el
c
h
ar
ac
t
er
is
t
ic
s
[
8
]
,
[
9
]
.
W
h
il
e
t
h
e
s
e
m
e
t
h
o
d
s
ar
e
w
i
d
el
y
u
s
ed
,
t
h
e
y
s
tr
u
g
g
le
to
ca
p
t
u
r
e
t
h
e
co
m
p
le
x
it
y
o
f
r
ea
l
d
r
i
v
i
n
g
i
n
m
e
tr
o
p
o
l
it
a
n
ar
e
a
s
.
T
h
e
r
e
s
e
ar
c
h
i
n
[
8
]
–
[
1
0
]
s
h
o
w
s
t
h
a
t
C
OP
E
R
T
o
u
tp
u
t
s
ca
n
v
ar
y
co
n
s
i
d
er
ab
l
y
a
n
d
s
o
m
e
t
i
m
e
s
o
v
e
r
p
r
ed
ic
t
C
O
2
e
m
i
s
s
io
n
s
a
t
lo
w
u
r
b
a
n
s
p
e
ed
s
,
i
n
d
i
ca
ti
n
g
t
h
e
n
e
ed
f
o
r
ad
j
u
s
t
m
e
n
t
u
s
i
n
g
lo
ca
l
d
ata
a
n
d
,
i
n
s
o
m
e
c
a
s
e
s
,
t
h
e
v
a
l
u
e
o
f
i
n
te
g
r
at
i
n
g
m
a
c
h
i
n
e
lea
r
n
i
n
g
te
c
h
n
iq
u
e
s
.
E
m
i
s
s
io
n
f
ac
to
r
s
u
n
iq
u
e
to
B
an
g
k
o
k
t
h
a
t
ar
e
g
e
n
er
at
ed
f
r
o
m
c
h
a
s
s
i
s
d
y
n
a
m
o
m
e
t
er
te
s
ti
n
g
h
a
v
e
a
ls
o
b
e
e
n
e
s
ta
b
li
s
h
ed
to
i
m
p
r
o
v
e
ac
c
u
r
ac
y
u
n
d
er
l
o
ca
l
d
r
i
v
i
n
g
co
n
d
i
tio
n
s
[
1
1
]
,
e
v
e
n
t
h
o
u
g
h
t
h
e
ir
p
r
a
ct
ic
al
a
p
p
l
ic
at
io
n
i
s
s
ti
ll
li
m
i
t
ed
b
e
ca
u
s
e
o
f
lab
o
r
ato
r
y
co
n
s
tr
ai
n
t
s
a
n
d
s
m
a
ll
s
a
m
p
l
e
q
u
a
n
t
i
ti
es
.
F
u
r
t
h
er
a
n
a
l
y
s
i
s
w
i
t
h
C
OP
E
R
T
f
o
r
t
h
e
B
a
n
g
k
o
k
Me
t
r
o
p
o
li
ta
n
R
e
g
i
o
n
p
r
o
v
id
e
s
co
n
t
e
x
t
f
o
r
u
r
b
a
n
e
m
i
s
s
io
n
s
[
1
2
]
;
n
e
v
er
t
h
e
le
s
s
,
t
h
e
s
e
m
e
t
h
o
d
o
lo
g
ie
s
s
t
i
ll
la
c
k
r
es
o
l
u
tio
n
a
t
t
h
e
v
e
h
i
cl
e
le
v
el.
I
n
ter
n
atio
n
all
y
,
m
ac
h
i
n
e
lear
n
in
g
m
o
d
els
h
av
e
d
e
m
o
n
s
tr
at
ed
th
e
ab
ilit
y
to
o
u
tp
er
f
o
r
m
tr
ad
itio
n
al
r
eg
r
ess
io
n
f
o
r
C
O
2
est
i
m
at
i
o
n
,
in
cl
u
d
in
g
ex
p
lai
n
ab
le
m
ac
h
in
e
lear
n
i
n
g
f
r
a
m
e
w
o
r
k
s
[
1
3
]
,
m
u
lti
-
m
o
d
el
en
s
e
m
b
les
co
m
p
ar
i
n
g
u
p
to
1
8
alg
o
r
ith
m
s
[
1
4
]
,
an
d
ap
p
licatio
n
s
h
i
g
h
l
ig
h
ti
n
g
t
h
e
s
u
p
er
io
r
ity
o
f
m
ac
h
i
n
e
lear
n
in
g
o
v
er
cla
s
s
ica
l
s
tat
is
t
ical
m
et
h
o
d
s
i
n
tr
an
s
p
o
r
t
-
e
m
is
s
io
n
an
al
y
s
is
[
1
5
]
.
Mo
r
eo
v
er
,
C
atB
o
o
s
t
-
b
ased
f
o
r
ec
asti
n
g
h
a
s
s
h
o
w
n
s
tr
o
n
g
ac
cu
r
ac
y
f
o
r
lig
h
t
-
d
u
t
y
v
e
h
icl
e
C
O
2
p
r
ed
ictio
n
an
d
r
o
b
u
s
tn
e
s
s
in
h
e
ter
o
g
en
eo
u
s
f
leets
s
i
m
i
lar
to
B
an
g
k
o
k
’
s
[
1
6
]
.
B
r
o
a
d
er
r
ev
ie
w
s
o
f
m
ac
h
i
n
e
lear
n
in
g
,
ar
tif
icial
n
e
u
r
al
n
et
w
o
r
k
s
,
an
d
d
ee
p
lear
n
in
g
m
eth
o
d
s
i
n
tr
an
s
p
o
r
t
-
e
m
is
s
io
n
s
m
o
d
eli
n
g
f
u
r
t
h
er
s
u
p
p
o
r
t
th
e
g
r
o
w
i
n
g
s
h
if
t
to
w
ar
d
d
ata
-
d
r
iv
e
n
ap
p
r
o
ac
h
es
[
1
7
]
.
Ho
w
e
v
er
,
m
o
s
t
e
x
i
s
ti
n
g
m
ac
h
i
n
e
lear
n
in
g
s
t
u
d
ies
r
el
y
o
n
ce
r
ti
f
icat
io
n
d
ataset
s
,
s
m
all
ex
p
er
i
m
e
n
tal
s
a
m
p
les,
o
r
v
e
h
icle
f
leets
f
r
o
m
n
o
n
-
A
s
ia
n
r
eg
io
n
s
,
li
m
it
in
g
t
h
eir
ap
p
licab
ilit
y
to
co
m
p
lex
,
m
i
x
ed
-
tec
h
n
o
lo
g
y
m
e
g
ac
itie
s
s
u
c
h
as B
an
g
k
o
k
.
T
ab
le
1
p
r
esen
ts
a
co
m
p
ar
ativ
e
o
v
er
v
ie
w
o
f
co
n
v
e
n
tio
n
a
l
e
m
is
s
io
n
est
i
m
a
tio
n
f
r
a
m
e
w
o
r
k
s
an
d
m
ac
h
in
e
lear
n
i
n
g
ap
p
r
o
ac
h
e
s
f
r
eq
u
e
n
tl
y
ap
p
lied
in
tr
a
n
s
p
o
r
tatio
n
-
r
elate
d
C
O
2
e
m
i
s
s
io
n
s
t
u
d
ies.
T
h
e
co
m
p
ar
is
o
n
h
i
g
h
li
g
h
ts
t
h
eir
m
et
h
o
d
o
lo
g
ical
s
tr
en
g
t
h
s
a
n
d
p
r
ac
tical
lim
i
tatio
n
s
.
I
t
also
ad
d
r
ess
es
th
eir
r
elev
an
ce
to
B
an
g
k
o
k
'
s
u
r
b
an
tr
af
f
ic
co
n
d
itio
n
s
an
d
h
eter
o
g
e
n
eo
u
s
v
e
h
icle
r
eg
i
s
tr
at
io
n
c
h
ar
ac
ter
is
tics
.
T
h
ese
co
llectiv
e
f
i
n
d
in
g
s
r
ev
ea
l
t
w
o
ess
e
n
tial
g
ap
s
.
First,
f
e
w
s
tu
d
ie
s
in
te
g
r
ate
d
etailed
v
eh
icle
attr
ib
u
tes,
s
u
c
h
a
s
m
o
d
el
y
e
ar
,
en
g
i
n
e
d
i
s
p
lace
m
e
n
t,
f
u
el
t
y
p
e,
a
n
d
E
u
r
o
ce
r
ti
f
icatio
n
,
w
ith
s
ta
n
d
ar
d
ized
e
m
is
s
io
n
f
ac
to
r
s
to
en
ab
le
v
eh
ic
le
-
lev
el
C
O
2
esti
m
atio
n
.
Seco
n
d
,
to
th
e
b
est
o
f
c
u
r
r
en
t
k
n
o
w
led
g
e
,
n
o
p
r
io
r
r
esear
ch
h
as
co
m
b
in
ed
E
P
A
lab
o
r
ato
r
y
-
te
s
ted
C
O
2
v
alu
e
s
,
C
OP
E
R
T
/Eu
r
o
r
eg
u
lato
r
y
f
ac
to
r
s
,
an
d
a
lar
g
e
-
s
ca
le
r
ea
l
-
w
o
r
ld
r
eg
is
tr
atio
n
d
ataset
w
it
h
i
n
a
u
n
if
ied
an
al
y
tical
a
n
d
m
ac
h
in
e
l
ea
r
n
in
g
f
r
a
m
e
w
o
r
k
f
o
r
B
an
g
k
o
k
.
T
o
ad
d
r
ess
th
ese
g
ap
s
,
t
h
is
s
tu
d
y
i
n
te
g
r
ates t
h
r
ee
co
m
p
le
m
e
n
tar
y
d
ataset
s
:
i)
L
ab
o
r
ato
r
y
-
b
ased
f
u
e
l
ec
o
n
o
m
y
an
d
C
O
2
m
ea
s
u
r
e
m
e
n
ts
f
r
o
m
th
e
U.
S.
E
n
v
ir
o
n
m
e
n
tal
P
r
o
tectio
n
Ag
e
n
c
y
(
E
P
A
)
.
ii)
A
co
m
p
r
eh
e
n
s
i
v
e
r
eg
is
tr
y
o
f
1
.
8
4
6
m
il
lio
n
v
e
h
icle
s
r
ec
o
r
d
ed
b
y
th
e
Dep
ar
t
m
e
n
t
o
f
L
an
d
T
r
an
s
p
o
r
t
(
DL
T
)
in
B
an
g
k
o
k
b
et
w
ee
n
2
0
2
3
an
d
2
0
2
5
.
iii)
C
OP
E
R
T
/Eu
r
o
4
–
6
em
i
s
s
io
n
f
ac
to
r
s
co
m
m
o
n
l
y
u
s
ed
i
n
in
ter
n
atio
n
al
e
m
i
s
s
io
n
-
in
v
e
n
to
r
y
s
y
s
te
m
s
.
T
h
ese
s
o
u
r
ce
s
ar
e
h
ar
m
o
n
ized
in
to
a
u
n
i
f
ied
v
eh
ic
le
-
le
v
el
a
n
al
y
tical
d
atase
t
to
d
ev
e
lo
p
an
d
ev
alu
a
te
th
r
ee
m
ac
h
i
n
e
lear
n
i
n
g
m
o
d
els
:
C
atB
o
o
s
t,
s
u
p
p
o
r
t
v
ec
to
r
r
eg
r
ess
io
n
(
SV
R
)
,
an
d
d
ee
p
n
eu
r
al
n
et
w
o
r
k
s
(
DNN)
.
T
h
e
m
o
d
els
ar
e
u
s
ed
to
esti
m
ate
r
ea
l
-
w
o
r
ld
C
O
2
e
m
is
s
io
n
s
at
th
e
in
d
i
v
id
u
al
-
ve
h
icle
lev
el
a
n
d
to
co
m
p
ar
e
p
r
ed
ictiv
e
p
er
f
o
r
m
a
n
ce
ac
r
o
s
s
en
s
e
m
b
le
-
b
ased
,
k
er
n
el
-
d
r
iv
e
n
,
an
d
d
ee
p
lea
r
n
in
g
ap
p
r
o
ac
h
es.
I
n
ad
d
itio
n
,
th
e
an
al
y
s
is
e
x
a
m
i
n
es
h
o
w
k
e
y
v
eh
icle
c
h
ar
ac
ter
i
s
tics
,
s
u
c
h
as
en
g
i
n
e
d
is
p
lace
m
en
t,
f
u
el
t
y
p
e,
a
n
d
E
u
r
o
ce
r
tif
icatio
n
,
s
h
ap
e
CO
2
-
e
m
is
s
io
n
b
e
h
av
io
r
,
th
er
eb
y
o
f
f
er
in
g
e
m
p
ir
ical
i
n
s
i
g
h
ts
t
h
at
m
a
y
in
f
o
r
m
tar
g
eted
r
eg
u
lato
r
y
m
ea
s
u
r
es a
n
d
f
u
t
u
r
e
tr
an
s
p
o
r
t
-
e
m
is
s
io
n
p
o
licies.
T
h
e
r
em
ai
n
d
er
o
f
t
h
e
ar
ticle
i
s
o
r
g
an
ized
as
f
o
llo
w
s
.
Sectio
n
2
p
r
o
v
id
es
a
d
etailed
d
escr
ip
tio
n
o
f
t
h
e
s
tu
d
y
ar
ea
,
th
e
d
ata
s
o
u
r
ce
s
u
s
ed
,
an
d
th
e
p
r
ep
r
o
ce
s
s
in
g
s
tep
s
ap
p
lied
to
p
r
ep
a
r
e
th
e
d
ata
f
o
r
an
aly
s
i
s
.
Sectio
n
3
d
escr
ib
es
th
e
ex
p
er
i
m
en
tal
w
o
r
k
f
lo
w
a
n
d
th
e
p
r
o
ce
d
u
r
es
f
o
llo
w
ed
d
u
r
in
g
m
o
d
el
tr
ain
in
g
.
Sectio
n
4
p
r
esen
ts
t
h
e
r
es
u
lt
s
,
in
cl
u
d
i
n
g
ass
es
s
m
en
t
s
o
f
v
ar
iab
le
i
m
p
o
r
tan
ce
,
m
ea
s
u
r
es
o
f
p
r
ed
ictio
n
ac
cu
r
ac
y
,
a
n
d
a
n
an
al
y
s
is
o
f
e
m
is
s
io
n
c
h
ar
ac
ter
is
tics
.
Sec
tio
n
5
co
n
cl
u
d
es th
e
p
ap
er
w
it
h
p
o
lic
y
co
n
s
id
er
atio
n
s
a
n
d
s
u
g
g
esti
o
n
s
f
o
r
d
ir
ec
tio
n
s
f
o
r
f
u
t
u
r
e
w
o
r
k
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J
A
d
v
A
p
p
l Sci
I
SS
N:
2252
-
8814
E
s
tima
tin
g
r
o
a
d
tr
a
n
s
p
o
r
t e
mis
s
io
n
s
in
Th
a
ila
n
d
u
s
in
g
d
a
ta
-
d
r
iven
mo
d
elin
g
…
(
P
h
a
c
h
a
ya
C
h
a
iw
ch
a
n
)
503
T
ab
le
1
.
C
o
m
p
ar
ativ
e
o
v
er
v
ie
w
o
f
C
O
2
e
m
is
s
io
n
e
s
ti
m
atio
n
m
et
h
o
d
s
A
p
p
r
o
a
c
h
/
mo
d
e
l
S
t
r
e
n
g
t
h
s
L
i
mi
t
a
t
i
o
n
s
R
e
l
e
v
a
n
c
e
t
o
B
a
n
g
k
o
k
c
o
n
t
e
x
t
K
e
y
r
e
f
e
r
e
n
c
e
s
T
h
a
i
l
o
c
a
l
e
mi
ssi
o
n
s
t
u
d
i
e
s
P
r
o
v
i
d
e
l
o
c
a
l
i
z
e
d
e
mi
ssi
o
n
f
a
c
t
o
r
s
a
n
d
r
e
a
l
-
w
o
r
l
d
c
o
n
t
e
x
t
u
a
l
i
n
s
i
g
h
t
s
L
i
mi
t
e
d
sca
l
a
b
i
l
i
t
y
f
o
r
l
a
r
g
e
f
l
e
e
t
-
l
e
v
e
l
p
r
e
d
i
c
t
i
o
n
a
n
d
d
y
n
a
mi
c
f
o
r
e
c
a
st
i
n
g
U
se
f
u
l
f
o
r
r
e
g
i
o
n
a
l
c
a
l
i
b
r
a
t
i
o
n
a
n
d
v
a
l
i
d
a
t
i
o
n
o
f
B
a
n
g
k
o
k
-
sp
e
c
i
f
i
c
e
mi
ss
i
o
n
c
h
a
r
a
c
t
e
r
i
st
i
c
s
N
i
l
r
i
t
e
t
a
l
.
[
2
]
,
[
3
]
,
S
i
r
i
t
h
i
a
n
e
t
a
l
.
[
1
1
]
,
K
u
so
n
e
t
a
l
.
[
1
2
]
C
O
P
ER
T
C
o
n
si
s
t
e
n
t
,
i
n
t
e
r
n
a
t
i
o
n
a
l
l
y
r
e
c
o
g
n
i
z
e
d
,
a
n
d
su
i
t
a
b
l
e
f
o
r
n
a
t
i
o
n
a
l
i
n
v
e
n
t
o
r
i
e
s
M
a
y
o
v
e
r
e
st
i
mat
e
e
mi
ssi
o
n
s
u
n
d
e
r
l
o
w
-
sp
e
e
d
c
o
n
g
e
st
i
o
n
a
n
d
st
o
p
-
a
n
d
-
g
o
t
r
a
f
f
i
c
c
o
n
d
i
t
i
o
n
s
B
a
n
g
k
o
k
’
s
h
i
g
h
l
y
c
o
n
g
e
st
e
d
u
r
b
a
n
t
r
a
f
f
i
c
r
e
q
u
i
r
e
s e
mp
i
r
i
c
a
l
a
d
j
u
s
t
me
n
t
a
n
d
l
o
c
a
l
i
z
e
d
c
a
l
i
b
r
a
t
i
o
n
L
e
j
r
i
e
t
a
l
.
[
8
]
,
D
e
y
e
t
a
l
.
[
9
]
,
C
O
P
ER
T
/
EEA
[
1
8
]
,
K
h
o
l
o
d
e
t
a
l
.
[
1
9
]
,
N
t
z
i
a
c
h
r
i
st
o
s
e
t
a
l
.
[
2
0
]
En
se
mb
l
e
mac
h
i
n
e
l
e
a
r
n
i
n
g
(
R
a
n
d
o
m
f
o
r
e
st
/
X
G
B
o
o
st
)
R
o
b
u
st
t
o
o
u
t
l
i
e
r
s;
e
f
f
e
c
t
i
v
e
f
o
r
n
o
n
l
i
n
e
a
r
t
a
b
u
l
a
r
p
r
e
d
i
c
t
i
o
n
a
n
d
f
e
a
t
u
r
e
i
n
t
e
r
a
c
t
i
o
n
mo
d
e
l
i
n
g
R
e
q
u
i
r
e
s c
a
r
e
f
u
l
e
n
c
o
d
i
n
g
a
n
d
m
a
y
su
f
f
e
r
f
r
o
m l
e
a
k
a
g
e
i
n
h
i
g
h
-
c
a
r
d
i
n
a
l
i
t
y
c
a
t
e
g
o
r
i
c
a
l
v
a
r
i
a
b
l
e
s
P
e
r
f
o
r
ms r
e
a
so
n
a
b
l
y
w
e
l
l
b
u
t
may
st
r
u
g
g
l
e
w
i
t
h
B
a
n
g
k
o
k
’
s
e
x
t
r
e
m
e
l
y
h
e
t
e
r
o
g
e
n
e
o
u
s D
L
T
c
a
t
e
g
o
r
i
c
a
l
st
r
u
c
t
u
r
e
G
u
r
c
a
n
[
1
4
]
,
Ç
ı
n
a
r
e
r
e
t
a
l
.
[
1
5
]
,
Z
h
o
u
e
t
a
l
.
[
1
6
]
S
V
R
S
t
r
o
n
g
n
o
n
l
i
n
e
a
r
a
p
p
r
o
x
i
mat
i
o
n
c
a
p
a
b
i
l
i
t
y
a
n
d
e
f
f
e
c
t
i
v
e
g
e
n
e
r
a
l
i
z
a
t
i
o
n
p
e
r
f
o
r
man
c
e
S
e
n
si
t
i
v
e
t
o
k
e
r
n
e
l
se
l
e
c
t
i
o
n
a
n
d
h
y
p
e
r
p
a
r
a
me
t
e
r
t
u
n
i
n
g
;
c
o
mp
u
t
a
t
i
o
n
a
l
l
y
e
x
p
e
n
si
v
e
f
o
r
v
e
r
y
l
a
r
g
e
d
a
t
a
se
t
s
B
a
n
g
k
o
k
’
s
d
a
t
a
se
t
(
~
1
.
8
M
)
p
o
se
s
c
o
mp
u
t
a
t
i
o
n
a
l
c
h
a
l
l
e
n
g
e
s
Ç
ı
n
a
r
e
r
e
t
a
l
.
[
1
5
]
,
N
a
t
a
r
a
j
a
n
e
t
a
l
.
[
2
1
]
O
p
e
n
v
e
h
i
c
l
e
sp
e
c
i
f
i
c
a
t
i
o
n
a
n
d
f
u
e
l
e
c
o
n
o
m
y
d
a
t
a
se
t
s
L
a
r
g
e
-
sca
l
e
st
a
n
d
a
r
d
i
z
e
d
v
e
h
i
c
l
e
i
n
f
o
r
mat
i
o
n
;
u
se
f
u
l
f
o
r
f
e
a
t
u
r
e
e
n
g
i
n
e
e
r
i
n
g
a
n
d
f
l
e
e
t
mat
c
h
i
n
g
M
a
y
l
a
c
k
l
o
c
a
l
i
z
e
d
d
r
i
v
i
n
g
b
e
h
a
v
i
o
r
a
n
d
o
p
e
r
a
t
i
o
n
a
l
v
a
r
i
a
b
i
l
i
t
y
S
u
p
p
o
r
t
s
i
n
t
e
g
r
a
t
i
o
n
w
i
t
h
B
a
n
g
k
o
k
D
L
T
r
e
g
i
st
r
a
t
i
o
n
a
t
t
r
i
b
u
t
e
s
a
n
d
f
l
e
e
t
h
e
t
e
r
o
g
e
n
e
i
t
y
O
p
e
n
d
a
t
a
so
f
t
d
a
t
a
se
t
[
2
2
]
,
D
L
T
d
a
t
a
se
t
[
2
3
]
Eu
r
o
4
–
6
c
e
r
t
i
f
i
c
a
t
i
o
n
st
a
n
d
a
r
d
s
En
a
b
l
e
s st
a
n
d
a
r
d
i
z
e
d
c
r
o
ss
-
man
u
f
a
c
t
u
r
e
r
c
o
mp
a
r
i
so
n
a
n
d
r
e
g
u
l
a
t
o
r
y
b
e
n
c
h
mar
k
i
n
g
L
a
b
o
r
a
t
o
r
y
-
b
a
se
d
v
a
l
u
e
s ma
y
d
i
f
f
e
r
su
b
s
t
a
n
t
i
a
l
l
y
f
r
o
m
r
e
a
l
-
w
o
r
l
d
u
r
b
a
n
e
mi
ssi
o
n
s
B
a
n
g
k
o
k
f
l
e
e
t
i
n
c
l
u
d
e
s
man
y
p
r
e
-
Eu
r
o
v
e
h
i
c
l
e
s a
n
d
m
i
x
e
d
t
e
c
h
n
o
l
o
g
i
e
s
Eu
r
o
p
e
a
n
C
o
mm
i
ssi
o
n
[
2
4
]
,
F
o
n
t
a
r
a
s
e
t
a
l
.
[
2
5
]
S
i
mu
l
a
t
i
o
n
mo
d
e
l
s
(
P
H
EM
,
M
O
V
ES)
C
a
p
t
u
r
e
s d
y
n
a
mi
c
b
e
h
a
v
i
o
r
;
h
i
g
h
r
e
a
l
i
sm
f
o
r
mi
c
r
o
-
sca
l
e
a
n
a
l
y
si
s
R
e
q
u
i
r
e
s h
i
g
h
-
r
e
so
l
u
t
i
o
n
sp
e
e
d
,
t
i
me
,
a
n
d
t
r
a
j
e
c
t
o
r
y
d
a
t
a
B
a
n
g
k
o
k
l
a
c
k
s
h
i
g
h
-
r
e
so
l
u
t
i
o
n
sp
e
e
d
p
r
o
f
i
l
e
d
a
t
a
L
e
j
r
i
e
t
a
l
.
[
8
]
,
F
o
n
t
a
r
a
s
e
t
a
l
.
[
2
5
]
L
i
n
e
a
r
/
t
r
a
d
i
t
i
o
n
a
l
r
e
g
r
e
ssi
o
n
S
i
mp
l
e
;
i
n
t
e
r
p
r
e
t
a
b
l
e
;
r
e
p
r
o
d
u
c
i
b
l
e
L
i
mi
t
e
d
a
b
i
l
i
t
y
t
o
mo
d
e
l
n
o
n
l
i
n
e
a
r
i
n
t
e
r
a
c
t
i
o
n
s
a
n
d
h
e
t
e
r
o
g
e
n
e
o
u
s fl
e
e
t
b
e
h
a
v
i
o
r
B
a
n
g
k
o
k
’
s
mi
x
e
d
v
e
h
i
c
l
e
t
e
c
h
n
o
l
o
g
i
e
s
a
n
d
n
o
n
l
i
n
e
a
r
t
r
a
f
f
i
c
c
o
n
d
i
t
i
o
n
s
r
e
d
u
c
e
p
r
e
d
i
c
t
i
v
e
c
a
p
a
b
i
l
i
t
y
N
a
t
a
r
a
j
a
n
e
t
a
l
.
[
2
1
]
G
r
a
d
i
e
n
t
b
o
o
st
i
n
g
f
o
r
c
a
t
e
g
o
r
i
c
a
l
d
a
t
a
(
C
a
t
B
o
o
st
)
Ef
f
i
c
i
e
n
t
l
y
h
a
n
d
l
e
s
h
i
g
h
-
c
a
r
d
i
n
a
l
i
t
y
c
a
t
e
g
o
r
i
c
a
l
v
a
r
i
a
b
l
e
s;
r
e
d
u
c
e
s t
a
r
g
e
t
l
e
a
k
a
g
e
;
st
r
o
n
g
p
e
r
f
o
r
man
c
e
o
n
h
e
t
e
r
o
g
e
n
e
o
u
s t
a
b
u
l
a
r
d
a
t
a
se
t
s
R
e
q
u
i
r
e
s
h
y
p
e
r
p
a
r
a
me
t
e
r
t
u
n
i
n
g
a
n
d
c
o
mp
u
t
a
t
i
o
n
a
l
o
p
t
i
m
i
z
a
t
i
o
n
f
o
r
l
a
r
g
e
-
sca
l
e
t
r
a
i
n
i
n
g
H
i
g
h
l
y
su
i
t
a
b
l
e
f
o
r
B
a
n
g
k
o
k
’
s
d
i
v
e
r
se
f
l
e
e
t
c
o
mp
o
si
t
i
o
n
a
n
d
l
a
r
g
e
c
a
t
e
g
o
r
i
c
a
l
f
e
a
t
u
r
e
sp
a
c
e
S
u
n
e
t
a
l
.
[
2
6
]
,
F
e
n
g
e
t
a
l
.
[
2
7
]
DNN
C
a
p
t
u
r
e
s h
i
e
r
a
r
c
h
i
c
a
l
n
o
n
l
i
n
e
a
r
r
e
l
a
t
i
o
n
s
h
i
p
s
a
n
d
c
o
m
p
l
e
x
f
e
a
t
u
r
e
i
n
t
e
r
a
c
t
i
o
n
s
S
e
n
si
t
i
v
e
t
o
o
v
e
r
f
i
t
t
i
n
g
;
r
e
q
u
i
r
e
s
n
o
r
mal
i
z
a
t
i
o
n
;
h
i
g
h
c
o
mp
u
t
a
t
i
o
n
a
l
c
o
s
t
Ef
f
e
c
t
i
v
e
f
o
r
l
a
r
g
e
-
sca
l
e
p
r
e
d
i
c
t
i
o
n
b
u
t
p
o
t
e
n
t
i
a
l
l
y
u
n
st
a
b
l
e
u
n
d
e
r
mi
x
e
d
I
C
E/
EV
/
h
y
b
r
i
d
f
l
e
e
t
d
i
s
t
r
i
b
u
t
i
o
n
s
G
h
o
r
b
a
l
e
t
a
l
.
[
2
8
]
,
K
o
mara
e
t
a
l
.
[
2
9
]
,
G
a
g
o
e
t
a
l
.
[
3
0
]
,
Al
-
N
e
f
a
i
e
a
n
d
A
l
d
h
y
a
n
i
[
3
1
]
2.
M
AT
E
RIAL
S AN
D
M
E
T
H
O
D
2
.
1
.
Study
a
re
a
a
nd
da
t
a
s
o
urce
s
T
h
is
s
tu
d
y
f
o
c
u
s
es
o
n
B
an
g
k
o
k
,
T
h
ailan
d
’
s
ca
p
ital
an
d
lar
g
e
s
t
cit
y
,
w
i
th
o
v
er
1
0
m
i
llio
n
p
o
p
u
latio
n
an
d
m
a
n
y
u
r
b
an
s
u
b
d
is
tr
icts
.
Fig
u
r
e
1
s
h
o
w
s
s
ig
n
i
f
ica
n
t
b
u
s
i
n
ess
zo
n
es,
h
i
g
h
-
tr
a
f
f
ic
r
o
u
tes,
an
d
m
i
x
ed
r
esid
en
tial
a
n
d
co
m
m
er
cia
l
d
is
tr
icts
i
n
th
e
s
p
ec
if
ied
ar
ea
.
T
h
e
d
iv
er
s
i
f
ied
v
e
h
icle
f
leet
i
n
B
an
g
k
o
k
,
in
c
lu
d
i
n
g
p
ass
en
g
e
r
v
e
h
icles,
b
u
s
e
s
,
h
e
av
y
-
d
u
t
y
tr
u
c
k
s
,
m
o
to
r
c
y
cles,
an
d
o
th
er
s
,
p
r
o
v
id
es
a
r
ea
lis
tic
en
v
ir
o
n
m
e
n
t
f
o
r
m
ea
s
u
r
in
g
C
O
2
e
m
is
s
io
n
s
.
T
h
e
in
co
r
p
o
r
atio
n
o
f
n
u
m
er
o
u
s
s
u
b
d
is
tr
ict
s
f
ac
ilit
a
tes
v
ar
iab
ilit
y
i
n
r
o
ad
in
f
r
astru
ct
u
r
e,
tr
af
f
ic
d
en
s
it
y
,
an
d
v
eh
icle
u
s
ag
e
p
atte
r
n
s
,
al
l
o
f
w
h
ic
h
ar
e
cr
it
ical
d
eter
m
in
an
ts
in
e
m
is
s
io
n
m
o
d
eli
n
g
.
T
o
esti
m
ate
C
O
2
em
is
s
io
n
s
f
r
o
m
v
e
h
icle
s
r
eg
is
t
er
ed
in
B
an
g
k
o
k
,
T
h
ailan
d
,
th
is
s
t
u
d
y
e
m
p
lo
y
s
th
r
ee
p
r
i
m
ar
y
d
ata
s
o
u
r
ce
s
to
co
n
s
tr
u
ct
m
ac
h
i
n
e
lear
n
i
n
g
m
o
d
els.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8814
I
n
t J
A
d
v
A
p
p
l Sci
,
Vo
l.
1
5
,
No
.
2
,
J
u
n
e
2
0
2
6
:
5
0
1
-
517
504
Fig
u
r
e
1
.
Stu
d
y
ar
ea
m
ap
2.
1
.
1
.
E
nv
iro
nm
ent
a
l
P
ro
t
ec
t
io
n Ag
ency
v
ehicle
em
is
s
io
ns
a
nd
f
uel e
co
no
m
y
da
t
a
s
et
T
h
e
in
itial
d
atase
t
w
as
s
o
u
r
ce
d
f
r
o
m
t
h
e
U.
S.
E
P
A
[
2
2
]
,
[
2
5
]
,
co
m
p
r
is
i
n
g
s
t
an
d
ar
d
ized
lab
o
r
ato
r
y
-
te
s
ted
d
ata
o
n
v
e
h
i
cle
f
u
el
ec
o
n
o
m
y
,
e
n
g
i
n
e
d
is
p
lace
m
en
t,
m
a
n
u
f
ac
t
u
r
er
s
p
ec
if
icatio
n
s
,
a
n
d
C
O
2
e
m
is
s
io
n
s
m
ea
s
u
r
ed
in
g
r
a
m
s
p
er
k
ilo
m
eter
(
g
/k
m
)
.
T
h
e
m
ea
s
u
r
e
m
en
ts
p
r
o
d
u
ce
d
at
th
e
E
P
A
Natio
n
al
Veh
icle
an
d
F
u
el
E
m
i
s
s
io
n
s
L
ab
o
r
ato
r
y
f
u
n
ctio
n
as
b
aseli
n
e
r
ef
er
en
ce
v
a
lu
e
s
.
T
h
ey
als
o
o
f
f
er
d
ep
en
d
ab
le
esti
m
ates o
f
v
e
h
icle
-
lev
e
l e
n
v
i
r
o
n
m
e
n
tal
p
er
f
o
r
m
a
n
ce
.
2
.
1
.
2
.
B
a
ng
k
o
k
v
ehicle
re
g
is
t
ra
t
io
n da
t
a
s
et
T
h
e
s
e
co
n
d
d
at
as
et
co
m
p
r
i
s
e
s
1
,
8
4
6
,
1
1
1
v
e
h
i
cl
e
r
e
g
i
s
tr
a
ti
o
n
r
ec
o
r
d
s
g
a
t
h
er
ed
b
y
T
h
a
il
an
d
’
s
D
L
T
f
r
o
m
J
a
n
u
ar
y
2
0
2
3
to
J
a
n
u
ar
y
2
0
2
5
[
2
3
]
.
T
h
e
d
at
as
et
e
n
c
o
m
p
a
s
s
e
s
v
e
h
i
cl
e
s
i
ze
cl
a
s
s
,
b
r
a
n
d
,
m
o
d
e
l,
f
u
e
l
t
y
p
e,
e
n
g
i
n
e
t
y
p
e,
a
n
d
e
m
i
s
s
io
n
-
co
n
tr
o
l
te
c
h
n
o
lo
g
i
e
s
.
T
h
er
eb
y
o
f
f
er
i
n
g
a
p
r
ec
i
s
e
r
ep
r
e
s
e
n
ta
t
io
n
o
f
B
an
g
k
o
k
’
s
o
n
-
r
o
ad
v
e
h
ic
le
f
le
et
f
o
r
l
o
ca
l
i
ze
d
e
m
i
s
s
io
n
m
o
d
el
i
n
g
a
n
d
t
r
e
n
d
a
n
a
l
y
s
i
s
.
2
.
1
.
3
.
E
m
is
s
io
n
s
t
a
nd
a
rds
a
nd
m
a
pp
ing
v
a
ria
bles
T
h
e
th
ir
d
d
ataset
in
clu
d
es
C
O
2
em
is
s
io
n
f
ac
to
r
s
o
b
tain
ed
f
r
o
m
C
OP
E
R
T
5
an
d
E
u
r
o
4
–
6
em
is
s
io
n
ce
r
tif
icatio
n
s
ta
n
d
ar
d
s
,
as
r
ep
o
r
ted
b
y
th
e
E
u
r
o
p
ea
n
E
n
v
ir
o
n
m
e
n
t
Ag
e
n
c
y
[
1
8
]
–
[
2
0
]
,
[
2
4
]
,
[
3
2
]
.
E
m
is
s
io
n
f
ac
to
r
s
w
er
e
allo
ca
ted
to
ea
c
h
r
eg
is
ter
ed
v
eh
ic
le
b
ased
o
n
c
r
iter
ia
in
cl
u
d
in
g
ce
r
ti
f
icatio
n
l
ev
el,
f
u
el
t
y
p
e,
a
n
d
en
g
i
n
e
tec
h
n
o
lo
g
y
.
I
n
t
h
e
ab
s
en
ce
o
f
d
ir
ec
t
e
m
is
s
io
n
n
u
m
b
er
s
f
r
o
m
E
P
A
r
ec
o
r
d
s
,
C
OP
E
R
T
-
b
ased
esti
m
ates
w
er
e
u
t
ilized
,
an
d
th
i
s
co
r
r
elatio
n
f
ac
ili
tates
a
s
ta
n
d
ar
d
ized
co
m
p
ar
is
o
n
o
f
m
ea
s
u
r
ed
an
d
r
eg
u
la
to
r
y
e
m
i
s
s
io
n
v
alu
e
s
an
d
s
u
p
p
o
r
ts
th
e
p
r
ed
ictio
n
o
f
C
O
2
e
m
i
s
s
io
n
s
ac
r
o
s
s
v
ar
io
u
s
v
e
h
icle
clas
s
es.
2
.
2
.
Da
t
a
prepro
ce
s
s
ing
R
ea
l
-
w
o
r
ld
v
eh
icle
e
m
is
s
io
n
s
d
if
f
er
b
y
3
0
%
–
4
0
%
f
r
o
m
lab
o
r
ato
r
y
te
s
t
v
alu
e
s
d
u
e
to
d
r
iv
in
g
b
eh
av
io
r
,
tr
af
f
ic
co
n
d
itio
n
s
,
an
d
en
v
ir
o
n
m
e
n
tal
f
ac
to
r
s
n
o
t
r
ep
r
esen
ted
in
co
n
tr
o
lled
test
in
g
[
2
5
]
.
T
h
e
i
m
p
le
m
en
ta
tio
n
o
f
a
s
tr
u
ctu
r
e
d
d
ata
p
r
et
r
ea
tm
e
n
t
an
d
in
te
g
r
atio
n
p
r
o
ce
d
u
r
e
w
a
s
n
ec
es
s
a
r
y
to
r
eso
lv
e
th
ese
d
is
p
ar
ities
.
I
t a
ls
o
en
h
a
n
ce
s
t
h
e
r
ea
lis
m
o
f
e
m
i
s
s
io
n
esti
m
ati
o
n
f
o
r
B
an
g
k
o
k
.
2
.
2
.
1
.
Da
t
a
clea
nin
g
a
nd
ha
rm
o
niza
t
io
n
A
ll
d
ata
s
ets
w
er
e
s
ta
n
d
ar
d
ized
to
en
s
u
r
e
co
n
s
i
s
te
n
c
y
i
n
v
ar
iab
le
n
a
m
e
s
,
u
n
it
s
,
an
d
ca
teg
o
r
ical
f
o
r
m
at
s
.
Ma
n
u
f
ac
t
u
r
er
an
d
m
o
d
el
f
ield
s
w
er
e
n
o
r
m
alize
d
ac
r
o
s
s
E
P
A
an
d
DL
T
r
ec
o
r
d
s
.
I
n
v
alid
,
d
u
p
licated
,
o
r
in
co
m
p
lete
e
n
tr
ies
w
er
e
r
e
m
o
v
ed
.
Ve
h
icles
th
at
co
u
ld
n
o
t
b
e
ass
o
ciate
d
w
it
h
e
m
is
s
io
n
f
ac
to
r
s
,
s
u
ch
a
s
n
o
n
-
m
o
to
r
ized
o
r
s
p
ec
ial
-
p
u
r
p
o
s
e
v
eh
ic
les
w
er
e
ex
cl
u
d
ed
f
r
o
m
t
h
e
a
n
al
y
s
is
.
2
.
2
.
2
.
I
nte
g
ra
t
io
n o
f
E
P
A,
DL
T
,
a
nd
CO
P
E
RT
/
E
uro
da
t
a
A
m
u
lti
-
s
ta
g
e
i
n
teg
r
atio
n
p
r
o
ce
d
u
r
e
w
as
ap
p
lied
to
co
m
b
in
e
t
h
e
t
h
r
ee
p
r
i
m
ar
y
d
ata
s
o
u
r
ce
s
:
i
)
E
P
A
lab
o
r
ato
r
y
-
test
ed
CO
2
e
m
is
s
io
n
s
an
d
v
eh
ic
le
s
p
ec
i
f
icatio
n
s
;
ii
)
DL
T
v
e
h
icle
r
eg
i
s
tr
atio
n
r
ec
o
r
d
s
f
o
r
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J
A
d
v
A
p
p
l Sci
I
SS
N:
2252
-
8814
E
s
tima
tin
g
r
o
a
d
tr
a
n
s
p
o
r
t e
mis
s
io
n
s
in
Th
a
ila
n
d
u
s
in
g
d
a
ta
-
d
r
iven
mo
d
elin
g
…
(
P
h
a
c
h
a
ya
C
h
a
iw
ch
a
n
)
505
B
an
g
k
o
k
f
r
o
m
J
an
u
ar
y
2
0
2
3
to
J
an
u
ar
y
2
0
2
5
;
an
d
iii
)
C
OP
E
R
T
an
d
E
u
r
o
4
–
6
em
is
s
io
n
f
ac
to
r
s
.
E
v
er
y
r
eg
is
ter
ed
v
e
h
icle
w
as
co
r
r
elate
d
w
it
h
E
P
A
in
f
o
r
m
atio
n
ac
co
r
d
in
g
to
b
r
an
d
,
m
o
d
el,
en
g
in
e
d
is
p
lace
m
e
n
t,
an
d
f
u
e
l
t
y
p
e
.
I
n
t
h
e
ab
s
e
n
ce
o
f
lab
o
r
ato
r
y
-
m
ea
s
u
r
ed
C
O
2
d
ata,
co
ef
f
icie
n
ts
d
er
i
v
ed
f
r
o
m
th
e
C
OP
E
R
T
m
et
h
o
d
o
lo
g
y
w
er
e
allo
ca
ted
b
ased
o
n
v
e
h
icle
cla
s
s
i
f
ic
atio
n
,
f
u
el
t
y
p
e,
an
d
E
u
r
o
ce
r
tif
icatio
n
.
T
h
is
m
et
h
o
d
o
lo
g
y
g
u
ar
a
n
teed
t
h
at
e
v
er
y
1
,
8
4
6
,
1
1
1
v
eh
icles
w
er
e
allo
ca
ted
a
d
esig
n
ated
C
O
2
r
ef
er
en
ce
v
al
u
e
f
o
r
th
e
co
n
s
tr
u
ctio
n
o
f
t
h
e
m
ac
h
i
n
e
lear
n
in
g
m
o
d
el.
F
r
o
m
th
e
f
u
l
l
1
.
8
4
6
m
i
ll
i
o
n
v
eh
i
cl
e
s
,
7
1
7
,
0
8
9
v
e
h
i
c
l
es
(
≈
3
8
.
8
%
)
c
o
u
l
d
b
e
m
a
t
ch
e
d
w
i
th
v
al
id
em
is
s
i
o
n
i
n
f
o
r
m
a
ti
o
n
(
E
P
A
o
r
C
O
PE
R
T
)
.
T
a
b
l
e
2
s
u
m
m
a
r
i
z
es
th
e
ch
a
r
a
c
t
e
r
is
t
i
cs
o
f
th
e
r
esu
l
ti
n
g
in
teg
r
a
t
e
d
d
a
t
as
e
t
.
T
h
is
e
m
p
ir
ical
s
u
m
m
ar
y
co
r
r
esp
o
n
d
s
w
ith
d
o
cu
m
e
n
ted
v
e
h
icle
tr
en
d
s
i
n
B
an
g
k
o
k
,
illu
s
tr
atin
g
a
f
leet
p
r
ed
o
m
in
a
n
tl
y
co
m
p
o
s
ed
o
f
g
aso
li
n
e
p
ass
en
g
er
v
eh
icles,
alo
n
g
s
id
e
an
in
cr
ea
s
i
n
g
p
r
e
v
alen
ce
o
f
b
atter
y
elec
tr
ic
v
eh
ic
les
(
B
E
V)
an
d
late
-
E
u
r
o
-
s
ta
n
d
ar
d
au
to
m
o
b
iles
.
T
ab
le
2
.
Su
m
m
ar
y
s
ta
tis
tic
s
o
f
th
e
in
te
g
r
ated
B
an
g
k
o
k
f
leet
d
ataset
V
a
r
i
a
b
l
e
V
a
l
u
e
/
d
i
s
t
r
i
b
u
t
i
o
n
T
o
t
a
l
mat
c
h
e
d
v
e
h
i
c
l
e
s
7
1
7
,
0
8
9
V
e
h
i
c
l
e
c
l
a
ss (%)
P
a
sse
n
g
e
r
c
a
r
s 9
4
.
5
%
,
mo
t
o
r
c
y
c
l
e
s
2
.
8
%,
p
i
c
k
u
p
s
/
S
U
V
s
1
.
8
%,
a
n
d
o
t
h
e
r
s
0
.
9
%
F
u
e
l
t
y
p
e
(
%)
G
a
so
l
i
n
e
9
3
.
7
%,
B
EV
3
.
7
%,
d
i
e
se
l
1
.
6
%,
a
n
d
h
y
b
r
i
d
/
P
H
EV
1
.
0
%
En
g
i
n
e
d
i
sp
l
a
c
e
me
n
t
(
c
c
)
M
e
a
n
±
S
D
=
2
,
0
5
7
±
5
,
7
1
4
Eu
r
o
st
a
n
d
a
r
d
d
i
s
t
r
i
b
u
t
i
o
n
Eu
r
o
≤
3
(
3
7
.
7
%),
E
u
r
o
4
(
3
8
.
3
%)
,
E
u
r
o
5
(
1
2
.
3
%)
,
a
n
d
E
u
r
o
6
(
1
1
.
4
%)
EPA
C
O
2
(
g
/
k
m)
M
e
a
n
±
S
D
=
3
2
2
.
2
±
5
5
.
3
R
e
g
i
st
r
a
t
i
o
n
y
e
a
r
r
a
n
g
e
1
9
9
0
–
2
0
2
5
2
.
2
.
3
.
F
ea
t
ure
eng
ineering
Sev
er
al
d
er
iv
ed
v
ar
iab
les
w
er
e
cr
ea
ted
to
im
p
r
o
v
e
p
r
ed
ictiv
e
ca
p
ac
ity
o
f
m
ac
h
i
n
e
lear
n
i
n
g
m
o
d
els.
E
n
g
i
n
e
d
is
p
lace
m
en
t
w
a
s
g
r
o
u
p
ed
in
to
p
r
ac
tical
ca
teg
o
r
ies
(
e.
g
.
,
<1
.
2
L
,
1
.
2
–
1
.
8
L
,
1
.
8
–
2
.
5
L
,
an
d
>2
.
5
L
)
to
ca
p
tu
r
e
n
o
n
li
n
ea
r
r
elatio
n
s
h
ip
s
b
et
w
ee
n
en
g
i
n
e
s
ize
a
n
d
C
O
2
o
u
tp
u
t.
E
u
r
o
e
m
i
s
s
i
o
n
s
ta
n
d
ar
d
s
w
er
e
ass
i
g
n
ed
b
ased
o
n
r
eg
u
lato
r
y
i
m
p
le
m
e
n
tatio
n
ti
m
eli
n
es
an
d
v
e
h
icle
m
an
u
f
ac
t
u
r
in
g
y
ea
r
s
,
w
h
ile
v
eh
icle
class
es,
li
k
e
p
ass
en
g
er
ca
r
s
,
p
ick
u
p
tr
u
ck
s
,
m
o
to
r
c
y
cles,
e
lectr
ic
v
eh
icle
s
,
an
d
s
p
o
r
t
-
u
ti
lit
y
v
e
h
icle
s
,
w
er
e
ca
teg
o
r
ized
to
r
ef
lect
f
u
n
ctio
n
al
d
if
f
er
en
ce
s
in
r
ea
l
-
w
o
r
ld
o
p
er
atio
n
.
Fu
el
s
y
s
te
m
co
n
f
ig
u
r
atio
n
s
w
er
e
f
u
r
t
h
er
class
i
f
ied
in
to
in
ter
n
al
co
m
b
u
s
tio
n
e
n
g
i
n
e
(
I
C
E
)
,
h
y
b
r
i
d
,
p
lu
g
-
i
n
h
y
b
r
id
(
P
HE
V)
,
an
d
B
E
V
ty
p
es.
A
s
tan
d
ar
d
ized
esti
m
ate
o
f
a
n
n
u
al
m
i
leag
e
w
a
s
ap
p
lied
ac
r
o
s
s
v
e
h
icle
s
to
s
u
p
p
o
r
t su
b
s
eq
u
en
t a
g
g
r
eg
a
tio
n
a
n
d
co
m
p
ar
ati
v
e
an
al
y
s
is
.
C
ate
g
o
r
ical
f
ea
tu
r
es
w
er
e
p
r
o
ce
s
s
ed
ac
co
r
d
in
g
to
th
e
r
eq
u
ir
e
m
en
ts
o
f
ea
ch
m
o
d
elin
g
tech
n
iq
u
e:
C
atB
o
o
s
t
w
as
ab
le
to
in
g
est
r
a
w
ca
teg
o
r
ical
v
ar
iab
les
d
ir
ec
tl
y
u
s
i
n
g
its
n
ativ
e
e
n
co
d
in
g
m
ec
h
a
n
i
s
m
s
,
w
h
er
ea
s
SV
R
an
d
DNN
r
eq
u
ir
ed
n
u
m
e
r
ical
tr
an
s
f
o
r
m
atio
n
v
ia
o
n
e
-
h
o
t
e
n
co
d
in
g
o
r
e
m
b
ed
d
in
g
r
ep
r
esen
tatio
n
s
.
2
.
2
.
4
.
Cro
s
s
-
v
a
lid
a
t
io
n str
a
t
eg
y
T
o
en
h
an
ce
r
o
b
u
s
tn
e
s
s
a
n
d
r
ed
u
ce
s
a
m
p
le
b
ias,
a
5
-
f
o
ld
cr
o
s
s
-
v
alid
atio
n
m
et
h
o
d
w
a
s
e
m
p
lo
y
ed
in
s
tead
o
f
a
s
tat
ic
tr
ain
–
v
a
lid
atio
n
–
test
d
iv
i
s
io
n
a
n
d
th
e
d
ataset
w
as
r
an
d
o
m
l
y
p
ar
titi
o
n
ed
in
to
f
i
v
e
eq
u
a
l
s
u
b
s
et
s
,
w
it
h
ea
ch
f
o
ld
u
ti
liz
ed
r
ep
ea
ted
ly
as
t
h
e
test
in
g
s
et
w
h
ile
th
e
r
e
m
ai
n
i
n
g
f
o
u
r
f
u
n
ctio
n
ed
as
th
e
tr
ain
i
n
g
s
et.
P
er
f
o
r
m
a
n
ce
m
etr
ics,
in
cl
u
d
i
n
g
co
e
f
f
icien
t
o
f
d
eter
m
i
n
atio
n
(
R
2
)
,
r
o
o
t
m
ea
n
s
q
u
ar
e
er
r
o
r
(
R
MSE
)
,
an
d
m
ea
n
a
b
s
o
l
u
te
er
r
o
r
(
MA
E
)
,
w
er
e
ca
lcu
lat
ed
f
o
r
ea
ch
r
ep
ea
t,
w
it
h
th
e
m
ea
n
a
n
d
s
tan
d
ar
d
d
ev
iatio
n
s
u
b
s
eq
u
en
t
l
y
r
ep
o
r
ted
.
Hen
ce
,
th
is
cr
o
s
s
-
v
alid
atio
n
s
tr
ate
g
y
e
n
s
u
r
es
t
h
at
m
o
d
el
ev
alu
a
tio
n
r
e
m
ain
s
s
tab
le
an
d
r
ep
r
esen
tati
v
e
ac
r
o
s
s
d
iv
er
s
e
v
e
h
icle
t
y
p
es,
f
u
el
ca
teg
o
r
ies,
an
d
e
m
is
s
io
n
le
v
els.
2
.
2
.
5
.
M
o
del
inp
ut
prepa
ra
t
io
n
T
h
e
m
ac
h
in
e
lear
n
i
n
g
m
o
d
els
w
er
e
tr
ain
ed
u
tili
zi
n
g
v
e
h
icle
-
le
v
el
p
r
ed
icto
r
s
,
in
clu
d
in
g
e
n
g
i
n
e
d
is
p
lace
m
e
n
t,
f
u
el
co
n
s
u
m
p
ti
o
n
,
f
u
el
t
y
p
e,
v
eh
icle
class
,
an
d
E
u
r
o
em
is
s
io
n
s
ta
n
d
ar
d
.
B
ef
o
r
e
f
in
al
m
o
d
el
b
u
ild
in
g
,
r
ec
u
r
s
i
v
e
f
ea
t
u
r
e
eli
m
i
n
atio
n
(
R
FE)
w
as
u
til
ized
to
ascer
tain
t
h
e
m
o
s
t
s
i
g
n
i
f
ic
an
t
p
r
ed
icto
r
s
an
d
m
i
n
i
m
ize
r
ed
u
n
d
a
n
c
y
i
n
t
h
e
f
ea
tu
r
e
s
et.
T
h
is
co
m
p
r
eh
e
n
s
i
v
e
p
r
ep
r
o
ce
s
s
in
g
p
ip
elin
e
,
in
co
r
p
o
r
atin
g
r
eg
u
lato
r
y
p
ar
am
eter
s
,
e
m
p
ir
icall
y
v
alid
ated
em
i
s
s
io
n
v
al
u
es,
a
n
d
ac
tu
al
f
lee
t
ch
ar
ac
ter
is
tic
s
,
y
ield
ed
a
h
ig
h
-
r
eso
l
u
tio
n
d
ataset
ap
p
r
o
p
r
iate
f
o
r
s
u
p
er
v
is
ed
lear
n
i
n
g
.
T
h
e
cu
r
ated
d
a
tas
et
o
f
f
er
s
a
s
o
lid
b
a
s
is
f
o
r
p
r
ec
is
el
y
m
o
d
elin
g
v
eh
ic
le
-
lev
el
C
O
2
e
m
is
s
io
n
s
u
n
d
er
v
ar
io
u
s
tr
a
f
f
ic
an
d
v
e
h
icl
e
co
n
d
itio
n
s
i
n
B
an
g
k
o
k
’
s
u
r
b
an
s
etti
n
g
.
2
.
3
.
M
o
delin
g
t
ec
hn
iqu
es
T
h
r
ee
s
u
p
er
v
is
ed
m
ac
h
i
n
e
lea
r
n
in
g
m
o
d
el
s
w
er
e
u
s
ed
to
esti
m
ate
v
eh
icle
-
lev
el
C
O
2
e
m
i
s
s
io
n
s
u
s
i
n
g
h
eter
o
g
e
n
eo
u
s
u
r
b
an
f
leet
d
a
ta:
DNN,
SV
R
,
an
d
C
atB
o
o
s
t.
T
h
ese
m
o
d
els
w
er
e
ch
o
s
e
n
to
o
v
er
co
m
e
t
h
e
s
h
o
r
tco
m
i
n
g
s
o
f
d
eter
m
i
n
i
s
tic
e
m
i
s
s
io
n
m
o
d
els
.
T
h
e
y
ad
d
r
ess
t
h
eir
co
m
p
le
m
e
n
t
ar
y
s
tr
en
g
th
s
,
w
h
ic
h
i
n
cl
u
d
e
h
an
d
li
n
g
ca
teg
o
r
ical
f
ea
tu
r
e
s
,
m
o
d
eli
n
g
n
o
n
li
n
ea
r
in
ter
ac
tio
n
s
,
an
d
ca
p
tu
r
i
n
g
h
i
g
h
-
co
m
p
le
x
it
y
p
atter
n
s
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8814
I
n
t J
A
d
v
A
p
p
l Sci
,
Vo
l.
1
5
,
No
.
2
,
J
u
n
e
2
0
2
6
:
5
0
1
-
517
506
2
.
3
.
1
.
Ca
t
B
o
o
s
t
C
atB
o
o
s
t
is
a
g
r
ad
ien
t
-
b
o
o
s
ti
n
g
m
et
h
o
d
s
p
ec
if
icall
y
d
esi
g
n
ed
f
o
r
tab
u
lar
d
atasets
th
at
ex
ce
ls
in
h
an
d
li
n
g
ca
teg
o
r
ical
v
ar
iab
le
s
w
i
th
o
u
t
n
ec
es
s
itati
n
g
co
n
s
id
er
ab
le
p
r
ep
ar
atio
n
.
I
t
b
u
ild
s
an
en
s
e
m
b
le
o
f
d
ec
is
io
n
tr
ee
s
in
s
u
cc
es
s
i
v
e
p
h
ases
,
w
it
h
ea
ch
tr
ee
r
ec
tify
i
n
g
th
e
r
esid
u
a
l
er
r
o
r
s
o
f
its
p
r
ed
ec
ess
o
r
s
,
s
o
ac
h
ie
v
i
n
g
e
lev
ated
f
o
r
ec
ast
ac
cu
r
ac
y
.
C
a
tB
o
o
s
t
is
t
h
u
s
id
ea
ll
y
s
u
ited
f
o
r
h
eter
o
g
e
n
eo
u
s
v
e
h
icle
d
ata
,
in
cl
u
d
in
g
b
r
an
d
,
m
o
d
el,
f
u
el
t
y
p
e,
E
u
r
o
e
m
is
s
io
n
s
ta
n
d
ar
d
,
an
d
m
an
u
f
ac
tu
r
e
y
ea
r
,
f
r
eq
u
e
n
tl
y
s
ee
n
i
n
e
m
is
s
io
n
m
o
d
eli
n
g
r
esear
ch
[
2
5
]
.
A
m
aj
o
r
b
en
ef
it
o
f
C
atB
o
o
s
t
its
o
r
d
er
e
d
tar
g
et
-
b
ased
en
co
d
in
g
,
an
ap
p
r
o
ac
h
th
at
p
r
ev
en
ts
tar
g
et
leak
ag
e
b
y
en
s
u
r
i
n
g
th
a
t c
ateg
o
r
ical
en
co
d
in
g
s
ar
e
g
en
er
ated
en
tire
l
y
f
r
o
m
i
n
f
o
r
m
atio
n
av
a
ilab
le
p
r
io
r
to
ea
ch
o
b
s
er
v
atio
n
an
d
th
is
m
e
th
o
d
s
u
b
s
ta
n
tiall
y
r
ed
u
ce
s
o
v
er
f
itt
in
g
,
a
p
r
ev
alen
t
p
r
o
b
lem
i
n
g
r
ad
ien
t
b
o
o
s
tin
g
w
h
e
n
ap
p
lied
to
h
ig
h
-
ca
r
d
in
al
it
y
c
ateg
o
r
ical
f
ea
t
u
r
es.
C
atB
o
o
s
t
ad
d
itio
n
all
y
i
n
co
r
p
o
r
a
tes
o
r
d
er
ed
b
o
o
s
tin
g
,
a
v
ar
ian
t
o
f
co
n
v
en
t
io
n
al
b
o
o
s
ti
n
g
t
h
at
s
ig
n
i
f
ica
n
tl
y
r
ed
u
ce
s
p
r
ed
ictio
n
b
ias.
T
h
e
C
atB
o
o
s
t
p
r
ed
ictio
n
ad
h
er
es
to
th
e
s
ta
n
d
ar
d
g
r
ad
ien
t b
o
o
s
tin
g
f
r
a
m
e
w
o
r
k
,
ar
ticu
lated
as
(
1
)
.
̂
(
)
=
0
(
)
+
∑
η
ℎ
(
)
=
1
(
1
)
W
h
er
e
̂
C
O
2
is
th
e
p
r
ed
icted
C
O
2
em
is
s
io
n
s
,
0
(
)
is
an
in
itial
m
o
d
el
p
r
ed
ictio
n
(
m
ea
n
o
f
C
O
2
),
(
)
is
th
e
p
r
ed
ictio
n
f
r
o
m
th
e
t
th
d
ec
is
io
n
tr
ee
,
η
is
th
e
lear
n
in
g
r
ate
,
an
d
is
th
e
to
tal
n
u
m
b
er
o
f
tr
ee
s
.
T
h
is
w
o
r
k
in
v
o
lv
ed
th
e
tu
n
in
g
o
f
C
atB
o
o
s
t
b
y
ad
j
u
s
tin
g
cr
itical
h
y
p
er
p
ar
am
eter
s
s
u
ch
as
tr
ee
d
ep
th
,
iter
atio
n
co
u
n
t,
an
d
lear
n
in
g
r
ate.
R
eg
u
lar
izatio
n
tech
n
iq
u
es
,
s
p
ec
if
ically
o
r
d
er
ed
b
o
o
s
tin
g
,
w
er
e
em
p
lo
y
ed
to
r
ed
u
ce
o
v
er
f
itti
n
g
.
T
h
ese
tech
n
iq
u
es
also
im
p
r
o
v
ed
g
en
er
aliza
b
ilit
y
ac
r
o
s
s
B
an
g
k
o
k
’
s
v
ar
ied
v
eh
icle
f
leet.
2
.
3
.
2
.
Su
pp
o
rt
v
ec
t
o
r
re
g
re
s
s
io
n
SVR
,
an
ex
ten
s
io
n
o
f
th
e
s
u
p
p
o
r
t
v
ec
to
r
m
ac
h
in
e
f
r
am
ew
o
r
k
,
w
as
in
clu
d
ed
as
a
b
aselin
e
r
eg
r
ess
io
n
m
o
d
el
to
b
en
ch
m
ar
k
tr
ad
itio
n
al
m
ac
h
in
e
lear
n
in
g
p
er
f
o
r
m
an
ce
ag
ain
s
t
en
s
em
b
le
(
C
atB
o
o
s
t)
an
d
d
ee
p
lear
n
in
g
(
DNN)
m
eth
o
d
s
.
Usi
n
g
th
e
r
ad
ial
b
asis
f
u
n
ctio
n
(
R
B
F)
k
er
n
el,
SVR
ca
n
ef
f
ec
tiv
ely
m
o
d
el
in
tr
icate
n
o
n
lin
ea
r
ass
o
ciatio
n
s
ac
r
o
s
s
v
eh
icle
-
lev
el
p
r
ed
icto
r
s
,
in
clu
d
in
g
en
g
in
e
s
ize,
f
u
el
ty
p
e,
E
u
r
o
s
tan
d
ar
d
,
an
d
f
u
el
co
n
s
u
m
p
tio
n
.
Ho
w
ev
er
,
SVR
is
k
n
o
w
n
to
b
e
s
en
s
itiv
e
to
h
y
p
er
p
ar
am
eter
s
elec
tio
n
an
d
m
ay
b
ec
o
m
e
co
m
p
u
tatio
n
ally
ex
p
en
s
iv
e
w
h
en
ap
p
lied
to
lar
g
e
-
s
ca
le
d
atasets
s
u
ch
as
B
an
g
k
o
k
’
s
1
.
8
4
m
illi
o
n
–
v
e
h
ic
le
r
eg
is
tr
y
.
T
h
e
g
o
al
o
f
SVR
is
to
lear
n
a
f
u
n
ctio
n
(
)
th
at
d
ev
iates
f
r
o
m
th
e
tr
u
e
tar
g
et
v
alu
es
b
y
n
o
m
o
r
e
th
an
a
to
ler
an
ce
m
ar
g
in
ε,
w
h
ile
m
ain
tain
in
g
th
e
f
u
n
ctio
n
as
f
lat
as
p
o
s
s
ib
le.
T
h
is
y
ield
s
th
e
ε
–
in
s
en
s
itiv
e
r
eg
r
ess
io
n
f
u
n
ctio
n
,
d
ef
in
ed
as
(
2
)
.
̂
(
)
=
⋅
+
(
2
)
W
h
er
e
̂
(
)
is
th
e
p
r
ed
icted
C
O
2
em
is
s
io
n
v
alu
e,
is
th
e
w
eig
h
t
v
ec
to
r
lear
n
ed
b
y
th
e
m
o
d
el,
is
th
e
f
ea
tu
r
e
v
ec
to
r
,
w
h
ich
m
ay
in
clu
d
e
f
u
el
ty
p
e,
E
u
r
o
s
tan
d
ar
d
,
m
iles
p
er
g
allo
n
(
MP
G
)
,
en
g
in
e
s
ize,
v
eh
icle
ty
p
e
,
an
d
o
th
er
m
in
u
s
s
u
b
s
tan
ce
s
,
an
d
is
th
e
b
ias
ter
m
.
SVR
u
s
es
th
e
ε
–
in
s
en
s
itiv
e
lo
s
s
f
u
n
ct
io
n
,
d
ef
in
ed
as
(
3
)
.
(
,
̂
)
=
{
0
,
|
−
̂
|
−
,
if
|
−
̂
|
≤
o
t
he
rwise
(
3
)
B
y
d
is
r
eg
ar
d
in
g
er
r
o
r
s
lo
w
er
th
an
,
th
is
f
o
r
m
u
latio
n
en
ab
les
th
e
m
o
d
el
to
co
n
ce
n
tr
ate
o
n
n
o
tab
le
d
ev
iatio
n
s
in
s
tead
o
f
in
d
is
cr
im
in
ately
m
in
im
izin
g
all
tr
ain
in
g
er
r
o
r
s
.
Hy
p
er
p
ar
am
eter
s
C
,
,
an
d
w
er
e
o
p
tim
ized
u
s
in
g
k
-
f
o
ld
cr
o
s
s
-
v
alid
atio
n
to
m
itig
ate
o
v
er
f
itti
n
g
an
d
en
s
u
r
e
r
o
b
u
s
t
p
er
f
o
r
m
an
ce
ac
r
o
s
s
d
iv
er
s
e
v
eh
icle
ca
teg
o
r
ies.
A
lth
o
u
g
h
SVR
p
r
o
v
id
es
a
r
ig
o
r
o
u
s
th
eo
r
etica
l
f
o
u
n
d
atio
n
,
its
s
ca
lab
ilit
y
lim
itatio
n
s
an
d
s
en
s
itiv
ity
to
k
er
n
el
p
ar
am
eter
s
m
ak
e
it
less
ef
f
ec
tiv
e
f
o
r
v
er
y
lar
g
e
h
eter
o
g
en
eo
u
s
d
atasets
,
w
h
ich
is
r
ef
lecte
d
in
its
co
m
p
ar
ativ
ely
lo
w
er
p
r
ed
ictiv
e
ac
cu
r
ac
y
in
th
is
s
tu
d
y
.
2
.
3
.
3
.
Dee
p
neura
l net
w
o
rk
A
f
ee
d
f
o
r
w
ar
d
DNN
w
a
s
i
m
p
le
m
e
n
ted
to
m
o
d
el
t
h
e
c
o
m
p
le
x
n
o
n
li
n
ea
r
i
n
ter
ac
tio
n
s
b
et
w
ee
n
v
eh
ic
le
ch
ar
ac
ter
is
t
ics
a
n
d
C
O
2
e
m
i
s
s
io
n
b
e
h
av
io
r
.
A
lt
h
o
u
g
h
D
NNs
ar
e
tr
ad
itio
n
all
y
ap
p
lied
to
class
if
icatio
n
p
r
o
b
lem
s
,
t
h
e
y
also
d
e
m
o
n
s
tr
ate
s
tr
o
n
g
p
er
f
o
r
m
a
n
ce
i
n
r
eg
r
ess
io
n
ta
s
k
s
.
I
n
t
h
ese
tas
k
s
,
th
e
o
b
j
ec
tiv
e
is
to
p
r
ed
ict
co
n
tin
u
o
u
s
n
u
m
er
ical
v
al
u
es
s
u
c
h
as
v
e
h
icle
-
le
v
el
C
O
2
e
m
is
s
io
n
s
d
er
iv
ed
f
r
o
m
e
n
g
i
n
e,
f
u
e
l,
an
d
r
eg
u
lato
r
y
at
tr
ib
u
tes.
T
h
e
f
o
r
w
ar
d
p
r
o
p
ag
atio
n
p
r
o
c
ess
f
o
r
ea
ch
la
y
er
l
in
t
h
e
n
et
wo
r
k
is
d
ef
i
n
ed
as
(
4
)
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J
A
d
v
A
p
p
l Sci
I
SS
N:
2252
-
8814
E
s
tima
tin
g
r
o
a
d
tr
a
n
s
p
o
r
t e
mis
s
io
n
s
in
Th
a
ila
n
d
u
s
in
g
d
a
ta
-
d
r
iven
mo
d
elin
g
…
(
P
h
a
c
h
a
ya
C
h
a
iw
ch
a
n
)
507
(
)
=
(
)
(
−
1
)
+
(
)
(
4
)
W
h
er
e
(
)
is
th
e
o
u
tp
u
t
o
f
lay
er
b
ef
o
r
e
ap
p
ly
in
g
th
e
ac
tiv
atio
n
f
u
n
ctio
n
(
p
r
e
-
ac
tiv
atio
n
v
alu
e)
,
(
)
r
ep
r
esen
ts
th
e
w
eig
h
t
v
ec
to
r
o
f
th
e
co
n
n
ec
tio
n
s
in
lay
er
,
(
−
1
)
is
th
e
o
u
tp
u
t
(
ac
tiv
atio
n
)
f
r
o
m
th
e
p
r
ev
io
u
s
lay
er
(
l
−1
)
,
an
d
(
)
d
en
o
tes th
e
b
ias ter
m
in
lay
er
.
A
ctiv
atio
n
is
th
en
co
m
p
u
ted
as
(
5
)
.
(
)
=
σ
(
(
)
)
(
5
)
T
h
e
D
NN
e
m
p
lo
y
ed
t
h
e
r
ec
ti
f
ied
li
n
ea
r
u
n
i
t
(
R
e
L
U)
a
s
th
e
a
ct
i
v
a
tio
n
f
u
n
c
tio
n
b
ec
au
s
e
o
f
i
t
s
s
t
ab
i
l
it
y
a
n
d
e
f
f
ic
ie
n
c
y
i
n
d
ee
p
ar
c
h
it
ec
t
u
r
es
.
M
o
d
e
l
o
p
t
i
m
i
za
tio
n
u
ti
li
ze
d
t
h
e
Ad
a
m
o
p
t
i
m
i
ze
r
,
w
h
i
c
h
ad
j
u
s
ts
l
ea
r
n
i
n
g
r
a
te
s
t
h
r
o
u
g
h
o
u
t
tr
a
i
n
i
n
g
to
e
n
h
a
n
ce
co
n
v
e
r
g
e
n
ce
s
p
ee
d
.
T
o
m
i
t
i
g
a
t
e
o
v
e
r
f
i
tt
i
n
g
,
p
ar
t
ic
u
lar
l
y
cr
u
ci
al
f
o
r
e
x
t
e
n
s
i
v
e
a
n
d
d
i
v
e
r
s
e
d
at
a
s
e
ts
,
r
e
g
u
lar
iz
at
io
n
ap
p
r
o
a
c
h
e
s
w
e
r
e
e
m
p
lo
y
e
d
,
i
n
c
l
u
d
i
n
g
d
r
o
p
o
u
t
la
y
e
r
s
a
n
d
ea
r
l
y
s
to
p
p
i
n
g
.
T
h
e
n
e
t
w
o
r
k
’
s
d
e
p
t
h
,
n
e
u
r
o
n
co
u
n
t
p
er
l
a
y
er
,
a
n
d
d
r
o
p
o
u
t
r
at
e
s
w
e
r
e
o
p
ti
m
i
ze
d
b
y
cr
o
s
s
-
v
a
lid
at
io
n
o
n
t
h
e
tr
a
i
n
i
n
g
d
a
ta
s
et
,
an
d
t
h
i
s
d
e
s
i
g
n
a
llo
w
s
t
h
e
DN
N
to
ca
p
t
u
r
e
h
ier
ar
c
h
ic
al
a
n
d
n
o
n
l
i
n
e
ar
c
o
r
r
el
at
io
n
s
a
m
o
n
g
v
e
h
i
cl
e
v
ar
i
ab
l
es
,
o
f
f
er
i
n
g
a
n
e
f
f
e
ct
i
v
e
co
m
p
le
m
e
n
t
ar
y
m
o
d
el
i
n
g
m
e
t
h
o
d
to
SV
R
a
n
d
C
atB
o
o
s
t.
2
.
3
.
4
.
H
y
perpa
ra
m
et
er
t
un
ing
a
nd
o
v
er
f
it
t
ing
m
it
ig
a
t
io
n
Al
l
t
h
r
e
e
m
a
c
h
i
n
e
l
ea
r
n
i
n
g
m
o
d
el
s
w
e
r
e
t
u
n
ed
u
ti
l
iz
i
n
g
a
5
-
f
o
ld
cr
o
s
s
-
v
al
id
a
ti
o
n
te
ch
n
iq
u
e
to
g
u
ar
a
n
te
e
co
n
s
i
s
te
n
t
g
e
n
er
a
l
iza
t
io
n
p
er
f
o
r
m
a
n
ce
t
h
r
o
u
g
h
o
u
t
B
a
n
g
k
o
k
’
s
d
i
v
er
s
e
v
e
h
i
cle
f
le
et
d
ata
s
et
,
wh
i
c
h
co
n
s
i
s
t
s
o
f
1
,
8
4
6
,
1
1
1
r
ec
o
r
d
s
.
C
r
o
s
s
-
v
a
lid
at
io
n
w
a
s
e
m
p
lo
y
e
d
to
r
ed
u
c
e
s
a
m
p
le
b
i
as
.
I
t
w
a
s
a
ls
o
u
s
e
d
to
o
b
t
ai
n
p
er
f
o
r
m
a
n
ce
e
s
ti
m
ate
s
t
h
a
t
ar
e
r
e
s
il
ie
n
t
a
cr
o
s
s
d
i
f
f
er
e
n
t
v
e
h
ic
le
c
at
e
g
o
r
ie
s
,
f
u
el
t
y
p
e
s
,
a
n
d
e
m
i
s
s
i
o
n
r
e
g
u
l
at
io
n
s
.
Fo
r
C
atB
o
o
s
t,
h
y
p
er
p
ar
a
m
ete
r
o
p
tim
izatio
n
f
o
cu
s
ed
o
n
tr
e
e
d
ep
th
(
6
–
1
0
)
,
l
ea
r
n
in
g
r
ate
(
0
.
0
1
–
0
.
3
)
,
n
u
m
b
er
o
f
b
o
o
s
tin
g
iter
atio
n
s
(
5
0
0
–
2
,
0
0
0
)
,
an
d
L
2
leaf
r
eg
u
lar
iza
tio
n
(
1
–
1
0
)
.
R
an
d
o
m
a
n
d
g
r
id
s
ea
r
ch
es
w
er
e
u
tili
ze
d
ac
co
r
d
in
g
to
th
e
s
en
s
i
tiv
it
y
o
f
th
e
p
ar
a
m
e
ter
s
.
T
h
e
o
r
d
er
ed
b
o
o
s
tin
g
m
ec
h
a
n
is
m
o
f
C
atB
o
o
s
t,
w
h
ic
h
in
h
ib
it
s
tar
g
et
lea
k
ag
e
d
u
r
in
g
ca
te
g
o
r
ical
f
ea
t
u
r
e
e
n
co
d
in
g
,
f
u
n
ct
io
n
ed
as
a
p
r
in
cip
al
m
et
h
o
d
f
o
r
m
iti
g
ati
n
g
o
v
er
f
itti
n
g
,
w
h
ile
ea
r
l
y
s
to
p
p
in
g
b
ased
o
n
v
al
i
d
atio
n
lo
s
s
w
a
s
e
m
p
lo
y
ed
to
av
er
t
s
u
p
er
f
l
u
o
u
s
iter
atio
n
s
o
n
t
h
e
ex
ten
s
i
v
e
d
ataset.
Fo
r
SVR
w
i
th
R
B
F
k
er
n
el,
r
e
g
u
lar
i
za
tio
n
p
ar
a
m
e
ter
C
(
0
.
1
–
1
0
0
)
,
k
er
n
el
co
ef
f
icie
n
t
(
1
e
-
4
–
1e
-
1
)
,
an
d
ε
-
i
n
s
e
n
s
iti
v
e
m
ar
g
i
n
(
0
.
0
0
1
–
0
.
1
)
w
er
e
t
u
n
ed
v
ia
e
x
h
au
s
ti
v
e
g
r
id
-
s
ea
r
ch
.
B
ec
au
s
e
SVR
e
x
h
ib
its
h
i
g
h
co
m
p
u
tatio
n
al
co
s
ts
o
n
d
atas
ets
ex
ce
ed
i
n
g
o
n
e
m
illi
o
n
s
a
m
p
les,
r
ep
r
esen
tat
iv
e
s
tr
ati
f
i
e
d
s
u
b
-
s
a
m
p
li
n
g
w
as
ap
p
lied
to
m
ai
n
tai
n
f
ea
s
ib
ilit
y
w
h
ile
p
r
eser
v
i
n
g
th
e
d
i
s
tr
ib
u
tio
n
o
f
k
e
y
v
eh
ic
le
attr
ib
u
te
s
(
f
u
el
t
y
p
e,
e
n
g
i
n
e
s
ize,
an
d
E
u
r
o
s
tan
d
ar
d
)
.
T
h
e
s
tan
d
ar
d
izatio
n
o
f
co
n
ti
n
u
o
u
s
p
r
ed
icto
r
s
an
d
th
e
in
tr
in
s
ic
m
ar
g
i
n
-
b
ased
lo
s
s
f
u
n
ctio
n
f
u
r
t
h
er
a
id
ed
in
th
e
r
e
g
u
la
tio
n
o
f
o
v
er
f
it
tin
g
.
T
h
e
DNN
’
s
ar
c
h
itect
u
r
e
w
as
o
p
tim
ized
b
y
v
ar
y
in
g
t
h
e
n
u
m
b
er
o
f
h
id
d
en
la
y
er
s
(
2
–
5
)
,
n
eu
r
o
n
s
p
er
lay
er
(
3
2
–
2
5
6
)
,
lear
n
in
g
r
ate
(
0
.
0
0
0
5
–
0
.
0
1
)
,
an
d
d
r
o
p
o
u
t
r
ates
(
0
.
1
–
0
.
4
)
.
T
h
e
m
o
d
el
e
m
p
lo
y
ed
R
e
L
U
ac
tiv
atio
n
w
i
th
A
d
a
m
o
p
ti
m
iz
atio
n
to
f
ac
ilit
ate
ef
f
icie
n
t
co
n
v
er
g
en
ce
i
n
th
e
h
i
g
h
-
d
i
m
e
n
s
io
n
al
f
ea
t
u
r
e
s
p
ac
e.
Ov
er
f
itti
n
g
w
a
s
alle
v
iated
w
i
th
th
e
i
m
p
le
m
e
n
tatio
n
o
f
d
r
o
p
o
u
t
la
y
er
s
,
L
2
w
ei
g
h
t
r
eg
u
l
ar
izatio
n
,
an
d
ea
r
l
y
h
alti
n
g
in
itiated
b
y
a
p
latea
u
i
n
v
alid
a
tio
n
p
er
f
o
r
m
an
ce
.
T
h
e
b
atch
s
ize
w
as
d
eter
m
i
n
ed
b
y
g
r
ap
h
ics
p
r
o
ce
s
s
in
g
u
n
it (
GP
U
)
m
e
m
o
r
y
li
m
itatio
n
s
to
i
m
p
r
o
v
e
tr
ai
n
i
n
g
e
f
f
icie
n
c
y
a
n
d
en
s
u
r
e
g
r
ad
i
en
t stab
ili
t
y
.
A
ll
h
y
p
er
p
ar
a
m
e
ter
r
an
g
e
s
wer
e
d
er
iv
ed
f
r
o
m
p
r
e
v
io
u
s
r
e
s
ea
r
ch
o
n
au
to
m
o
b
ile
e
m
i
s
s
io
n
m
o
d
elin
g
an
d
r
ef
in
ed
th
r
o
u
g
h
t
h
e
f
ir
s
t
p
ilo
t
ex
p
er
im
en
ts
co
n
d
u
cted
o
n
s
u
b
s
ets
o
f
th
e
B
an
g
k
o
k
d
ataset.
T
h
e
tu
n
in
g
tech
n
iq
u
es
w
er
e
cr
u
cia
l
f
o
r
m
ain
tai
n
i
n
g
t
h
e
m
o
d
els
’
g
e
n
er
a
lizab
ilit
y
ac
r
o
s
s
v
ar
io
u
s
v
eh
ic
le
ca
teg
o
r
ies
.
T
h
e
y
also
en
s
u
r
ed
r
o
b
u
s
tn
e
s
s
ac
r
o
s
s
r
ea
lis
tic
e
m
i
s
s
io
n
s
ce
n
ar
io
s
in
m
etr
o
p
o
litan
m
e
g
ac
itie
s
.
3.
E
XP
E
R
I
M
E
NT
A
L
T
h
is
s
ec
tio
n
p
r
ese
n
ts
th
e
e
x
p
er
im
e
n
tal
w
o
r
k
f
lo
w
u
s
ed
to
co
n
s
tr
u
c
t,
ca
lib
r
ate,
an
d
ev
alu
ate
t
h
e
m
ac
h
in
e
lear
n
in
g
m
o
d
els
f
o
r
esti
m
ati
n
g
v
e
h
icle
-
le
v
el
C
O
2
e
m
is
s
io
n
s
in
B
an
g
k
o
k
.
T
h
e
w
o
r
k
f
lo
w
co
n
s
i
s
ts
o
f
s
ev
e
n
s
eq
u
en
t
ial
p
h
ase
s
:
d
ata
ac
q
u
is
it
io
n
,
d
ataset
i
n
t
eg
r
atio
n
,
p
r
ep
r
o
ce
s
s
in
g
,
f
ea
tu
r
e
en
g
i
n
ee
r
in
g
,
m
o
d
el
d
e
v
elo
p
m
e
n
t,
h
y
p
er
p
ar
a
m
eter
o
p
ti
m
izat
io
n
,
an
d
f
i
n
a
l
ev
al
u
atio
n
,
d
esi
g
n
ed
to
en
s
u
r
e
m
et
h
o
d
o
lo
g
ical
tr
a
n
s
p
ar
en
c
y
a
n
d
f
u
ll
r
ep
r
o
d
u
cib
ilit
y
.
Fig
u
r
e
2
s
u
m
m
ar
iz
es
th
e
co
m
p
lete
an
al
y
tical
p
ip
elin
e
ad
o
p
ted
in
th
is
s
t
u
d
y
.
3
.
1
.
Da
t
a
prepa
ra
t
io
n a
nd
da
t
a
inte
g
ra
t
io
n
T
o
d
ev
elo
p
a
co
m
p
r
eh
en
s
i
v
e
v
eh
icle
-
lev
el
d
atab
ase
f
o
r
m
o
d
elin
g
C
O
2
e
m
is
s
io
n
s
f
r
o
m
t
h
e
1
,
846
,
1
1
1
v
eh
icle
s
r
eg
is
ter
ed
in
B
an
g
k
o
k
,
t
h
r
ee
co
m
p
le
m
e
n
tar
y
d
ataset
s
w
er
e
i
n
te
g
r
ated
:
i
)
lab
o
r
ato
r
y
-
test
ed
f
u
e
l
ec
o
n
o
m
y
a
n
d
C
O
2
e
m
i
s
s
io
n
m
ea
s
u
r
e
m
e
n
ts
p
r
o
v
id
e
d
b
y
th
e
U.
S.
E
P
A
[
1
1
]
,
ii
)
n
atio
n
al
v
e
h
icle
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8814
I
n
t J
A
d
v
A
p
p
l Sci
,
Vo
l.
1
5
,
No
.
2
,
J
u
n
e
2
0
2
6
:
5
0
1
-
517
508
r
eg
is
tr
atio
n
r
ec
o
r
d
s
ac
q
u
ir
ed
f
r
o
m
th
e
D
L
T
b
et
w
ee
n
J
an
u
ar
y
202
3
an
d
J
an
u
ar
y
2025
,
an
d
[
12
]
,
an
d
iii)
CO
2
e
m
i
s
s
io
n
f
ac
to
r
s
o
b
tain
ed
f
r
o
m
t
h
e
C
OP
E
R
T
5
m
o
d
el
an
d
th
e
E
u
r
o
4
-
6
r
eg
u
lato
r
y
s
ta
n
d
ar
d
s
is
s
u
ed
b
y
t
h
e
E
u
r
o
p
ea
n
E
n
v
ir
o
n
m
en
t
A
g
e
n
c
y
[
1
8
]
–
[
2
0
]
,
[
2
4
]
,
[
3
2
]
.
T
h
e
E
P
A
d
ataset
s
u
p
p
lied
s
tan
d
ar
d
ized
p
o
llu
tio
n
b
en
ch
m
ar
k
s
,
en
g
in
e
s
p
ec
s
,
a
n
d
f
u
el
ec
o
n
o
m
y
r
ati
n
g
s
.
Me
an
w
h
ile,
th
e
D
L
T
d
ataset
p
r
esen
ted
a
d
etailed
d
ep
ictio
n
o
f
B
an
g
k
o
k
’
s
ac
t
u
al
v
eh
icle
f
leet
m
a
k
eu
p
,
e
n
co
m
p
as
s
i
n
g
b
r
an
d
,
m
o
d
el,
s
ize
cla
s
s
,
e
n
g
in
e
co
n
f
i
g
u
r
atio
n
,
an
d
f
u
el
t
y
p
e.
T
o
m
er
g
e
th
ese
d
atase
ts
,
E
P
A
a
n
d
DL
T
en
tr
ies
w
er
e
h
ar
m
o
n
ized
u
s
i
n
g
s
h
ar
ed
id
en
ti
f
i
er
s
s
u
c
h
as
m
an
u
f
ac
t
u
r
er
,
m
o
d
el
n
a
m
e,
en
g
i
n
e
d
is
p
lace
m
en
t,
an
d
f
u
el
s
y
s
te
m
attr
ib
u
te
s
.
Fu
zz
y
-
m
atc
h
in
g
p
r
o
ce
d
u
r
es
w
er
e
e
m
p
lo
y
ed
to
r
eso
lv
e
v
ar
ia
tio
n
s
i
n
n
a
m
i
n
g
co
n
v
e
n
tio
n
s
ac
r
o
s
s
s
o
u
r
ce
s
a
n
d
to
id
en
tify
t
h
e
clo
s
es
t
E
P
A
eq
u
iv
ale
n
t
f
o
r
ea
ch
r
e
g
is
ter
ed
v
eh
icle.
W
h
e
n
a
m
a
tch
to
E
PA
d
ata
co
u
ld
n
o
t
b
e
estab
li
s
h
e
d
,
C
OP
E
R
T
-
b
ased
CO
2
co
ef
f
icie
n
ts
w
er
e
as
s
i
g
n
e
d
ac
co
r
d
in
g
to
v
eh
icle
cla
s
s
,
f
u
el
t
y
p
e,
an
d
E
u
r
o
ce
r
tif
icatio
n
lev
el.
Af
ter
i
n
teg
r
ati
n
g
all
s
o
u
r
ce
s
,
r
ec
o
r
d
s
lack
in
g
b
o
t
h
E
P
A
v
a
lu
es
a
n
d
C
OP
E
R
T
-
ass
ig
n
ab
le
e
m
is
s
io
n
f
ac
to
r
s
w
er
e
r
e
m
o
v
ed
.
T
h
is
f
ilter
i
n
g
s
tep
p
r
o
d
u
ce
d
7
1
7
,
0
8
9
v
eh
icles
(
3
8
.
8
%
o
f
th
e
o
r
ig
in
a
l
f
leet)
w
i
th
co
m
p
lete
a
n
d
v
al
id
attr
ib
u
tes
r
eq
u
ir
ed
f
o
r
C
O
2
esti
m
atio
n
an
d
m
ac
h
i
n
e
lear
n
i
n
g
m
o
d
eli
n
g
.
T
h
is
i
n
te
g
r
ated
d
ataset
f
o
r
m
ed
a
s
tab
le,
in
ter
n
all
y
co
n
s
is
te
n
t b
asi
s
f
o
r
s
u
b
s
e
q
u
en
t a
n
al
y
s
is
.
Fig
u
r
e
2
.
W
o
r
k
f
lo
w
of
CO
2
e
m
is
s
io
n
s
esti
m
atio
n
an
d
m
ac
h
in
e
lear
n
i
n
g
an
al
y
s
is
3
.
2
.
E
m
is
s
io
ns
e
s
t
i
m
a
t
io
n
met
ho
do
lo
g
y
CO
2
e
m
i
s
s
io
n
s
f
o
r
ea
ch
v
e
h
i
cle
w
er
e
esti
m
ated
u
s
i
n
g
i
n
t
er
n
atio
n
all
y
r
ec
o
g
n
ized
m
e
th
o
d
o
lo
g
ies,
in
cl
u
d
in
g
t
h
e
I
P
C
C
2
0
0
6
g
u
id
elin
es
f
o
r
Natio
n
al
Gr
ee
n
h
o
u
s
e
Gas I
n
v
e
n
to
r
ies
[
3
3
]
,
2
0
1
9
r
ef
i
n
e
m
e
n
t
[
3
4
]
,
an
d
th
e
E
u
r
o
p
ea
n
E
n
v
ir
o
n
m
en
t
Ag
e
n
c
y
’
s
C
OP
E
R
T
5
f
r
a
m
e
w
o
r
k
[
2
0
]
an
d
th
e
E
P
A
d
at
aset
’
s
f
u
el
ec
o
n
o
m
y
n
u
m
b
er
s
w
er
e
n
o
r
m
alize
d
b
y
co
n
v
er
ti
n
g
MP
G
to
liter
s
p
er
k
ilo
m
e
ter
(
L
/k
m
)
to
g
u
ar
an
te
e
co
n
s
is
te
n
c
y
w
it
h
e
m
is
s
io
n
f
ac
to
r
es
ti
m
ates
u
s
e
d
in
b
o
th
t
h
e
I
P
C
C
an
d
C
OP
E
R
T
ap
p
r
o
ac
h
es.
Fo
r
ea
ch
v
e
h
icle
r
ec
o
r
d
,
an
n
u
a
l
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J
A
d
v
A
p
p
l Sci
I
SS
N:
2252
-
8814
E
s
tima
tin
g
r
o
a
d
tr
a
n
s
p
o
r
t e
mis
s
io
n
s
in
Th
a
ila
n
d
u
s
in
g
d
a
ta
-
d
r
iven
mo
d
elin
g
…
(
P
h
a
c
h
a
ya
C
h
a
iw
ch
a
n
)
509
CO
2
e
m
i
s
s
io
n
s
w
er
e
co
m
p
u
t
ed
as
th
e
p
r
o
d
u
ct
o
f
f
u
el
co
n
s
u
m
p
tio
n
,
th
e
co
r
r
esp
o
n
d
in
g
e
m
is
s
io
n
f
ac
to
r
(
ex
p
r
ess
ed
in
g
r
a
m
s
o
f
C
O
2
p
er
liter
o
f
f
u
el)
,
an
d
an
ass
u
m
ed
an
n
u
al
d
r
iv
in
g
d
is
ta
n
ce
o
f
1
5
,
0
0
0
k
m
,
co
n
s
is
ten
t
w
i
th
r
e
g
io
n
al
tr
a
n
s
p
o
r
tatio
n
s
tu
d
ie
s
.
T
h
e
g
en
er
al
e
s
ti
m
atio
n
f
o
r
m
u
la
is
e
x
p
r
ess
ed
as
(
6
).
2
=
(
)
×
(
/
)
×
15000
(
6
)
W
h
er
e
f
u
el
co
n
s
u
m
p
tio
n
d
en
o
tes
t
h
e
s
tan
d
ar
d
ized
f
u
el
-
u
s
e
r
ate,
an
d
t
h
e
e
m
i
s
s
io
n
f
a
cto
r
r
ep
r
esen
ts
t
h
e
a
m
o
u
n
t
o
f
C
O
2
e
m
itted
p
er
lit
er
o
f
f
u
e
l
co
m
b
u
s
ted
.
E
m
i
s
s
io
n
f
ac
to
r
s
v
ar
y
ac
co
r
d
in
g
to
f
u
el
t
y
p
e,
co
m
b
u
s
tio
n
ch
ar
ac
ter
is
tic
s
,
an
d
E
u
r
o
e
m
i
s
s
io
n
ce
r
ti
f
icatio
n
le
v
els
(
E
u
r
o
4
–
6
)
,
f
o
llo
w
in
g
E
u
r
o
p
ea
n
E
n
v
ir
o
n
m
en
t
Ag
en
c
y
C
OP
E
R
T
f
r
am
e
w
o
r
k
an
d
s
ta
n
d
ar
d
ized
g
r
ee
n
h
o
u
s
e
g
as
e
m
i
s
s
i
o
n
f
ac
to
r
r
ef
er
en
ce
s
[
1
0
]
,
[
2
0
]
,
[
3
5]
–
[
3
7
]
.
T
h
is
h
y
b
r
id
esti
m
atio
n
ap
p
r
o
ac
h
,
in
teg
r
at
in
g
e
m
p
ir
ical
E
P
A
-
b
as
ed
em
i
s
s
io
n
b
en
ch
m
ar
k
s
w
it
h
C
OP
E
R
T
-
d
er
iv
ed
r
eg
u
lato
r
y
co
e
f
f
icien
ts
,
en
s
u
r
es
s
cie
n
ti
f
icall
y
co
n
s
is
te
n
t
an
d
p
o
licy
-
r
ele
v
an
t
C
O
2
esti
m
a
tes
f
o
r
all
7
1
7
,
0
8
9
r
eg
is
ter
ed
v
eh
icles,
t
h
er
eb
y
p
r
o
v
id
in
g
a
r
o
b
u
s
t
f
o
u
n
d
atio
n
f
o
r
s
u
p
er
v
i
s
ed
m
ac
h
i
n
e
lear
n
in
g
an
a
l
y
s
is
.
3
.
3
.
Da
t
a
clea
nin
g
a
nd
prepro
ce
s
s
ing
A
s
tr
u
ct
u
r
ed
p
r
ep
r
o
ce
s
s
in
g
p
i
p
elin
e
w
as i
m
p
le
m
en
ted
to
en
s
u
r
e
t
h
at
i
n
teg
r
ated
E
P
A
–
D
L
T
–
C
OP
E
R
T
d
ataset
w
as
co
n
s
is
ten
t,
co
m
p
lete,
an
d
an
al
y
ticall
y
r
elia
b
le.
Af
ter
m
er
g
i
n
g
t
h
e
t
h
r
e
e
s
o
u
r
ce
s
,
s
e
v
er
al
in
co
n
s
is
te
n
cie
s
an
d
m
i
s
s
i
n
g
v
alu
es
w
er
e
id
en
ti
f
ied
,
p
ar
ticu
l
ar
l
y
i
n
f
u
el
-
ec
o
n
o
m
y
m
ea
s
u
r
em
en
ts
,
C
O
2
v
al
u
es
,
an
d
f
u
el
-
t
y
p
e
c
lass
if
icatio
n
s
.
Fu
r
t
h
er
m
o
r
e,
r
ec
o
r
d
s
lack
in
g
es
s
en
tia
l
attr
ib
u
te
s
r
eq
u
ir
ed
f
o
r
e
m
is
s
io
n
esti
m
atio
n
—
s
u
ch
a
s
f
u
el
t
y
p
e
,
en
g
i
n
e
d
is
p
lace
m
e
n
t,
o
r
E
PA
/
C
OP
E
R
T
e
m
is
s
io
n
v
al
u
es
—
w
er
e
r
e
m
o
v
ed
to
av
o
id
s
y
s
te
m
atic
b
ias.
Fo
r
v
ar
iab
les
w
it
h
p
ar
tial
m
is
s
in
g
n
es
s
(
e.
g
.
,
MP
G
f
o
r
ce
r
tain
m
o
d
el
-
y
ea
r
g
r
o
u
p
s
)
,
w
it
h
i
n
-
ca
te
g
o
r
y
m
ed
ia
n
i
m
p
u
t
atio
n
w
as a
p
p
lied
to
p
r
eser
v
e
d
is
tr
ib
u
tio
n
al
c
h
ar
ac
ter
is
tic
s
.
Du
e
to
th
e
ab
s
e
n
ce
o
f
s
p
ec
i
f
ic
E
u
r
o
e
m
is
s
io
n
cr
iter
ia
i
n
DL
T
r
eg
is
tr
atio
n
s
,
ea
ch
v
e
h
icle
w
a
s
ca
teg
o
r
ized
in
to
E
u
r
o
3
–
6
b
ased
o
n
its
r
eg
i
s
tr
atio
n
y
ea
r
,
ad
h
er
in
g
to
le
g
i
s
lati
v
e
ad
o
p
tio
n
ti
m
etab
les
a
s
s
h
o
w
n
in
T
ab
le
3
.
T
h
is
r
u
le
-
b
ased
class
i
f
icatio
n
g
u
ar
an
teed
alig
n
m
e
n
t
w
it
h
C
OP
E
R
T
e
m
is
s
i
o
n
f
ac
to
r
s
.
I
t
also
en
ab
led
u
n
i
f
o
r
m
f
ac
to
r
allo
ca
t
io
n
ac
r
o
s
s
v
e
h
icle
clas
s
i
f
icatio
n
s
.
T
o
im
p
r
o
v
e
p
r
ed
ictiv
e
p
er
f
o
r
m
a
n
ce
,
s
e
v
er
al
en
g
in
ee
r
ed
f
ea
tu
r
e
s
w
er
e
d
ev
elo
p
ed
,
in
cl
u
d
in
g
d
is
p
lace
m
e
n
t
ca
te
g
o
r
ies,
f
u
el
-
s
y
s
te
m
clas
s
if
icatio
n
s
(
I
C
E
,
h
y
b
r
id
,
P
HE
V,
an
d
B
E
V)
,
E
u
r
o
ce
r
tific
atio
n
lev
els,
a
n
d
d
er
iv
ed
e
m
is
s
io
n
m
etr
ics.
N
u
m
er
ical
v
ar
iab
les
w
er
e
s
tan
d
ar
d
ized
f
o
r
SV
R
a
n
d
DNN
to
e
n
h
an
c
e
co
n
v
er
g
e
n
ce
s
tab
il
it
y
,
w
h
er
ea
s
C
atB
o
o
s
t
u
ti
lized
r
a
w
v
alu
e
s
o
w
i
n
g
to
its
r
esil
ie
n
ce
w
it
h
v
ar
ied
tab
u
lar
d
ata.
T
h
is
m
u
lt
i
-
s
tep
p
r
ep
ar
atio
n
p
ip
elin
e
g
en
er
ated
a
r
ef
in
ed
,
s
tan
d
ar
d
ized
d
ataset
a
p
p
r
o
p
r
iate
f
o
r
ex
ten
s
i
v
e
s
u
p
er
v
i
s
ed
lear
n
in
g
,
w
h
ile
en
s
u
r
i
n
g
m
et
h
o
d
o
lo
g
ical
tr
an
s
p
a
r
en
c
y
an
d
r
ep
ea
tab
ilit
y
w
it
h
i
n
B
an
g
k
o
k
'
s
d
iv
er
s
e
u
r
b
an
f
lee
t.
T
ab
le
3
.
E
u
r
o
em
is
s
io
n
s
ta
n
d
a
r
d
R
e
g
i
st
r
a
t
i
o
n
y
e
a
r
Eu
r
o
st
a
n
d
a
r
d
a
ssi
g
n
e
d
≤
2
0
0
9
Eu
r
o
3
o
r
u
n
d
e
r
2
0
1
0
–
2
0
1
4
Eu
r
o
4
2
0
1
5
–
2
0
1
9
Eu
r
o
5
≥
2
0
2
0
Eu
r
o
6
3
.
4
.
M
a
chine le
a
rning
t
ec
hn
iqu
es
T
h
r
ee
s
u
p
er
v
is
ed
m
ac
h
in
e
lea
r
n
in
g
al
g
o
r
ith
m
s
,
C
atB
o
o
s
t,
S
VR
,
an
d
DNN,
w
er
e
u
tili
ze
d
t
o
f
o
r
ec
ast
v
eh
ic
le
-
lev
el
C
O
2
e
m
i
s
s
io
n
s
f
r
o
m
B
an
g
k
o
k
’
s
d
i
v
er
s
e
u
r
b
an
f
leet.
T
h
ese
m
o
d
els
w
e
r
e
ch
o
s
en
f
o
r
th
eir
co
m
p
le
m
e
n
t
ar
y
ab
ilit
ies
to
ca
p
tu
r
e
n
o
n
li
n
ea
r
p
atter
n
s
,
m
a
n
ag
e
m
i
x
ed
ca
teg
o
r
ical
a
n
d
n
u
m
er
ical
s
tr
u
ct
u
r
es,
a
n
d
s
i
m
u
late
h
i
g
h
-
d
i
m
en
s
io
n
a
l
co
n
n
ec
tio
n
s
t
y
p
ical
l
y
f
o
u
n
d
in
tr
an
s
p
o
r
tatio
n
-
e
m
i
s
s
io
n
d
ata.
P
r
ev
io
u
s
s
tu
d
ie
s
h
av
e
es
tab
lis
h
ed
th
e
ap
p
r
o
p
r
iaten
es
s
o
f
C
atB
o
o
s
t
f
o
r
tab
u
lar
d
atasets
w
it
h
h
i
g
h
-
ca
r
d
in
alit
y
ca
teg
o
r
ical
f
ea
t
u
r
es
[
2
6
]
,
[
2
7
]
,
th
e
ef
f
icac
y
o
f
DNN
i
n
ca
p
t
u
r
in
g
m
u
lti
s
ca
le
n
o
n
l
in
ea
r
i
n
ter
ac
tio
n
s
i
n
en
v
ir
o
n
m
e
n
tal
an
d
en
g
i
n
ee
r
i
n
g
co
n
te
x
ts
[
2
8
]
–
[
3
0
]
,
[
3
8
]
,
an
d
th
e
t
h
eo
r
etica
l
r
es
ilien
ce
o
f
SV
R
a
s
a
r
o
b
u
s
t
n
o
n
li
n
ea
r
b
en
c
h
m
ar
k
f
o
r
r
eg
r
ess
io
n
tas
k
s
[
3
1
]
.
C
atB
o
o
s
t,
a
g
r
ad
ien
t
-
b
o
o
s
ti
n
g
d
ec
is
io
n
-
tr
ee
alg
o
r
it
h
m
,
w
a
s
o
p
ti
m
ized
w
ith
o
r
d
er
ed
b
o
o
s
tin
g
an
d
tar
g
et
-
b
a
s
ed
en
co
d
in
g
to
m
iti
g
ate
tar
g
et
leak
ag
e
an
d
d
i
m
i
n
is
h
p
r
ed
ictio
n
b
ias.
I
ts
in
h
er
en
t
ca
p
ac
it
y
to
p
r
o
ce
s
s
u
n
r
e
f
i
n
ed
ca
teg
o
r
ical
v
ar
iab
les,
s
u
c
h
a
s
v
e
h
icle
b
r
an
d
,
m
o
d
el,
f
u
e
l
t
y
p
e,
a
n
d
E
u
r
o
ce
r
tif
i
ca
tio
n
,
n
e
g
ated
th
e
n
ec
es
s
it
y
f
o
r
s
u
b
s
ta
n
tial
en
co
d
in
g
,
r
en
d
er
in
g
it
p
ar
ticu
lar
l
y
ap
p
r
o
p
r
iate
f
o
r
T
h
ailan
d
’
s
v
ar
ied
an
d
h
ig
h
-
ca
r
d
in
ali
t
y
v
eh
icle
r
eg
is
t
er
.
C
r
u
cial
h
y
p
er
p
ar
a
m
e
ter
s
,
s
u
ch
a
s
tr
ee
d
ep
th
,
lear
n
in
g
r
at
e,
an
d
th
e
n
u
m
b
er
o
f
b
o
o
s
tin
g
iter
at
io
n
s
,
w
er
e
m
o
d
if
ied
to
i
m
p
r
o
v
e
g
en
er
aliza
t
io
n
p
er
f
o
r
m
an
ce
a
m
o
n
g
f
leet
s
u
b
g
r
o
u
p
s
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8814
I
n
t J
A
d
v
A
p
p
l Sci
,
Vo
l.
1
5
,
No
.
2
,
J
u
n
e
2
0
2
6
:
5
0
1
-
517
510
SVR
w
it
h
R
B
F
k
er
n
el
s
er
v
ed
as
a
th
eo
r
etica
ll
y
g
r
o
u
n
d
ed
n
o
n
li
n
ea
r
b
aselin
e.
T
h
e
m
et
h
o
d
is
ca
p
ab
le
o
f
m
o
d
eli
n
g
co
m
p
lex
r
ela
tio
n
s
h
ip
s
a
m
o
n
g
e
n
g
i
n
e
d
is
p
lace
m
en
t,
E
u
r
o
s
ta
n
d
ar
d
,
f
u
e
l
t
y
p
e,
an
d
f
u
el
-
ec
o
n
o
m
y
v
ar
iab
les.
B
ec
au
s
e
SV
R
b
ec
o
m
es
co
m
p
u
tatio
n
all
y
p
r
o
h
ib
it
iv
e
o
n
d
ataset
s
e
x
ce
ed
in
g
o
n
e
m
il
lio
n
s
a
m
p
le
s
,
h
y
p
er
p
ar
a
m
eter
tu
n
i
n
g
w
as
co
n
d
u
cted
o
n
s
tr
atif
ied
s
u
b
s
a
m
p
les
t
h
at
p
r
eser
v
ed
th
e
d
is
tr
ib
u
tio
n
a
l
ch
ar
ac
ter
is
tic
s
o
f
th
e
o
v
er
all
f
leet.
Mo
d
el
p
ar
am
eter
s
(
C
,
γ
,
an
d
ε)
w
er
e
o
p
ti
m
ized
to
b
ala
n
ce
f
le
x
ib
ilit
y
a
n
d
r
esil
ien
ce
a
g
ai
n
s
t o
v
er
f
itti
n
g
.
A
f
u
ll
y
co
n
n
ec
ted
f
ee
d
f
o
r
w
ar
d
DNN
ar
ch
itect
u
r
e
w
as
i
m
p
l
e
m
en
ted
to
ca
p
tu
r
e
m
u
lt
ilev
e
l
n
o
n
li
n
ea
r
r
elatio
n
s
h
ip
s
in
h
er
en
t
in
r
ea
l
-
w
o
r
ld
e
m
is
s
io
n
b
eh
a
v
io
r
.
T
h
e
n
et
w
o
r
k
e
m
p
lo
y
ed
R
e
L
U
a
cti
v
atio
n
,
th
e
A
d
a
m
o
p
ti
m
izer
,
an
d
r
eg
u
lar
i
za
tio
n
tech
n
iq
u
e
s
,
in
cl
u
d
in
g
d
r
o
p
o
u
t,
b
atch
n
o
r
m
aliza
tio
n
,
an
d
ea
r
l
y
s
to
p
p
in
g
,
to
s
tab
ilize
tr
ain
i
n
g
a
n
d
co
n
tr
o
l
o
v
er
f
itti
n
g
.
A
r
c
h
itect
u
r
al
p
ar
a
m
eter
s
s
u
ch
a
s
th
e
n
u
m
b
e
r
o
f
h
id
d
en
la
y
er
s
,
u
n
i
ts
p
er
lay
er
,
lear
n
in
g
r
ate,
an
d
d
r
o
p
o
u
t
r
atio
s
w
er
e
t
u
n
ed
th
r
o
u
g
h
cr
o
s
s
-
v
alid
atio
n
to
en
s
u
r
e
r
o
b
u
s
t
p
r
ed
ictiv
e
p
er
f
o
r
m
a
n
ce
.
A
ll
th
r
ee
m
o
d
els
w
er
e
tr
ai
n
ed
u
til
izin
g
t
h
e
id
en
tical
ly
d
esi
g
n
ed
f
ea
tu
r
e
s
et
d
eli
n
ea
ted
in
s
u
b
-
s
ec
tio
n
3
.
3
to
g
u
ar
an
tee
eq
u
it
y
in
co
m
p
ar
is
o
n
as
s
es
s
m
en
t
.
T
h
e
ev
al
u
atio
n
o
f
t
h
e
m
o
d
el
p
er
f
o
r
m
a
n
ce
u
tili
ze
d
R
M
SE,
MA
E
,
an
d
R
2
.
I
t
f
ac
ilit
ated
a
th
o
r
o
u
g
h
a
n
d
in
ter
n
all
y
co
h
er
en
t
co
m
p
ar
is
o
n
a
m
o
n
g
en
s
e
m
b
le
-
b
a
s
ed
,
k
er
n
el
-
b
ased
,
an
d
d
ee
p
lear
n
in
g
m
et
h
o
d
o
lo
g
ies.
3
.
5
.
H
y
perpa
ra
m
et
er
o
ptim
iza
t
io
n
T
o
im
p
r
o
v
e
f
o
r
e
ca
s
t
ac
cu
r
a
c
y
an
d
g
u
a
r
an
t
ee
s
t
r
o
n
g
g
en
e
r
aliz
ati
o
n
t
o
B
an
g
k
o
k
’
s
d
iv
er
s
e
f
le
et
,
all
th
r
e
e
m
o
d
e
ls
u
n
d
er
w
en
t
ex
t
en
s
iv
e
h
y
p
er
p
a
r
am
eter
o
p
tim
izat
io
n
.
A
d
d
it
io
n
ally
,
a
5
-
f
o
ld
cr
o
s
s
-
v
al
id
ati
o
n
m
eth
o
d
w
as
u
s
ed
f
o
r
all
m
o
d
el
s
t
o
m
in
i
m
ize
s
am
p
lin
g
b
i
as
.
I
t
w
as
c
r
u
ci
al
f
o
r
g
en
e
r
a
tin
g
ac
c
u
r
at
e
p
er
f
o
r
m
an
ce
esti
m
ates g
iv
en
th
e
d
at
aset’
s
d
i
v
er
s
e
d
is
t
r
i
b
u
ti
o
n
s
o
f
f
u
el
ty
p
e
s
,
en
g
in
e
d
is
p
l
ac
em
en
ts
,
an
d
E
u
r
o
s
tan
d
a
r
d
s
.
C
atB
o
o
s
t
p
ar
am
eter
s
w
er
e
o
p
tim
ized
u
s
in
g
a
co
m
b
in
atio
n
o
f
r
an
d
o
m
ized
s
ea
r
ch
an
d
g
r
id
r
ef
in
em
en
t.
Key
p
ar
am
eter
s
in
c
lu
d
ed
tr
ee
d
ep
th
(
6
–
1
0
)
,
lear
n
in
g
r
ate
(
0
.
0
1
–
0
.
3
)
,
n
u
m
b
er
o
f
b
o
o
s
tin
g
iter
atio
n
s
(
5
0
0
–
2
,
0
0
0
)
,
an
d
L
2
leaf
r
eg
u
lar
izatio
n
(
1
–
1
0
)
.
Or
d
er
ed
b
o
o
s
tin
g
,
w
h
ich
p
r
ev
en
ts
tar
g
et
leak
ag
e
w
h
en
en
co
d
in
g
h
ig
h
-
ca
r
d
in
ality
ca
teg
o
r
ical
v
ar
iab
les,
s
er
v
ed
as
an
ess
en
tial
r
e
g
u
lar
izatio
n
m
ec
h
an
is
m
.
E
ar
ly
s
to
p
p
in
g
w
as
ap
p
lied
to
av
o
id
u
n
n
ec
ess
ar
y
b
o
o
s
tin
g
iter
atio
n
s
o
n
ce
v
alid
atio
n
lo
s
s
p
latea
u
ed
.
Op
tim
izatio
n
w
as
p
er
f
o
r
m
ed
o
n
s
tr
atif
ied
s
u
b
s
am
p
les
th
at
p
r
eser
v
ed
th
e
d
is
tr
ib
u
tio
n
o
f
f
u
el
ty
p
e,
en
g
in
e
d
is
p
lace
m
en
t,
an
d
E
u
r
o
s
tan
d
ar
d
,
b
ased
o
n
th
e
co
m
p
u
tatio
n
al
ex
p
en
s
e
o
f
SVR
o
n
d
atasets
.
A
th
o
r
o
u
g
h
g
r
id
s
ea
r
ch
o
p
tim
ized
th
e
p
en
alty
p
ar
am
eter
C
(
0
.
1
–
1
0
0
)
,
k
er
n
el
co
ef
f
icien
t
γ
(
1
e
-
4
–
1e
-
1
)
,
an
d
ε
-
in
s
en
s
itiv
e
m
ar
g
in
ε
(
0
.
0
0
1
–
0
.
1
)
.
T
h
e
s
tan
d
ar
d
izatio
n
o
f
co
n
tin
u
o
u
s
f
ea
tu
r
es
en
s
u
r
ed
n
u
m
er
ical
s
tab
ilit
y
,
w
h
ile
th
e
m
ar
g
in
-
b
ased
lo
s
s
f
u
n
ctio
n
w
as
r
eg
u
lar
ized
to
m
itig
ate
o
v
er
f
itti
n
g
.
T
h
e
h
y
p
er
p
ar
am
eter
s
o
f
th
e
DNN
,
n
etw
o
r
k
d
ep
th
(
2
–
5
lay
er
s
)
,
an
d
n
u
m
b
er
o
f
n
eu
r
o
n
s
p
er
lay
er
(
3
2
–
2
5
6
)
,
lear
n
in
g
r
ate
(
0
.
0
0
0
5
–
0
.
0
1
)
,
an
d
d
r
o
p
o
u
t
r
atio
s
(
0
.
1
–
0
.
4
)
,
w
er
e
o
p
tim
ized
b
y
iter
ativ
e
v
alid
atio
n
.
T
h
e
R
eL
U
ac
tiv
atio
n
f
u
n
ctio
n
an
d
th
e
A
d
am
o
p
tim
izer
f
ac
ilit
ated
ef
f
icien
t
co
n
v
er
g
en
ce
in
h
ig
h
-
d
im
en
s
io
n
al
f
ea
tu
r
e
s
p
ac
es.
R
eg
u
lar
izatio
n
m
eth
o
d
s
,
in
clu
d
in
g
d
r
o
p
o
u
t,
L
2
w
eig
h
t
d
ec
ay
,
an
d
ea
r
ly
s
to
p
p
in
g
,
w
er
e
em
p
lo
y
ed
to
en
h
an
ce
tr
ain
in
g
s
tab
ilit
y
an
d
m
itig
ate
o
v
er
f
itti
n
g
,
w
h
ich
is
p
ar
ticu
lar
ly
cr
u
cial
d
u
e
to
th
e
d
ataset
’
s
s
ca
le
an
d
d
iv
e
r
s
ity
.
A
ll
h
y
p
er
p
ar
am
eter
co
n
f
ig
u
r
atio
n
s
w
er
e
in
itially
ass
ess
ed
o
n
s
u
b
s
ets
o
f
th
e
in
teg
r
ated
d
atab
ase
to
en
s
u
r
e
co
m
p
u
tatio
n
al
f
ea
s
ib
ilit
y
.
F
in
al
tu
n
in
g
p
ar
am
eter
s
w
er
e
ch
o
s
en
b
ased
o
n
th
eir
ca
p
ac
ity
to
b
alan
ce
m
o
d
el
ac
cu
r
ac
y
.
T
h
ey
also
en
s
u
r
ed
co
n
v
er
g
en
ce
s
tab
ilit
y
an
d
g
en
er
aliza
tio
n
ac
r
o
s
s
v
eh
icle
ca
teg
o
r
ies,
f
u
el
s
y
s
tem
s
,
an
d
em
is
s
io
n
lev
els
.
3
.
6
.
M
o
del
ev
a
lua
t
io
n a
nd
s
t
a
t
is
t
ica
l t
esting
T
h
e
p
r
ed
ictiv
e
p
er
f
o
r
m
an
ce
o
f
th
e
th
r
ee
m
ac
h
in
e
lear
n
in
g
m
o
d
els
w
as
ev
alu
ated
u
s
in
g
a
co
m
p
r
eh
en
s
iv
e
ev
alu
atio
n
f
r
am
e
w
o
r
k
to
en
s
u
r
e
a
r
ig
o
r
o
u
s
an
d
in
ter
n
ally
co
n
s
is
ten
t
co
m
p
ar
is
o
n
.
T
h
e
R
2
,
MA
E
,
an
d
R
MSE
,
th
r
ee
w
id
ely
u
tili
ze
d
m
etr
ics
f
o
r
ass
ess
in
g
r
eg
r
ess
io
n
q
u
ality
in
en
v
ir
o
n
m
en
tal
an
d
tr
an
s
p
o
r
tatio
n
m
o
d
elin
g
,
w
er
e
em
p
lo
y
ed
to
q
u
an
tif
y
th
e
m
o
d
els'
ac
cu
r
ac
y
.
T
h
e
m
etr
ics
w
er
e
co
m
p
u
ted
u
n
d
er
co
n
s
is
ten
t
ex
p
er
im
en
tal
co
n
d
itio
n
s
to
g
u
ar
an
tee
co
m
p
ar
ab
ilit
y
am
o
n
g
C
atB
o
o
s
t,
SVR
,
an
d
DNN.
A
n
o
n
p
ar
am
etr
ic
b
o
o
ts
tr
ap
ap
p
r
o
ac
h
w
ith
1
,
0
0
0
r
esam
p
les
w
as
em
p
lo
y
ed
f
o
r
ea
ch
s
tatis
tic
to
ad
d
r
ess
s
am
p
lin
g
v
ar
iab
ilit
y
w
ith
in
th
e
ex
ten
s
iv
e
an
d
d
iv
er
s
e
f
leet
d
ataset.
T
h
e
r
esu
ltan
t
9
5
%
co
n
f
id
en
ce
in
ter
v
als
q
u
an
tif
y
m
o
d
el
u
n
ce
r
tain
ty
an
d
d
im
in
is
h
r
elian
ce
o
n
a
s
in
g
u
lar
tr
ain
-
test
d
iv
is
io
n
,
th
u
s
en
h
an
cin
g
th
e
r
esil
ien
ce
an
d
r
eliab
ilit
y
o
f
p
er
f
o
r
m
an
ce
ass
ess
m
en
ts
.
T
h
is
m
eth
o
d
o
lo
g
y
co
n
f
o
r
m
s
to
b
est
p
r
ac
tices
in
d
ata
-
d
r
iv
en
em
is
s
io
n
m
o
d
elin
g
,
w
h
er
ein
ass
ess
in
g
u
n
ce
r
tain
ty
is
eq
u
ally
cr
u
cial
as p
r
esen
tin
g
p
o
in
t v
alu
es.
I
n
ad
d
itio
n
to
th
e
n
u
m
er
ical
r
esu
lts
,
th
e
s
tu
d
y
also
ex
am
in
ed
h
o
w
ea
ch
m
o
d
el
b
eh
av
ed
ac
r
o
s
s
d
if
f
er
en
t
p
ar
ts
o
f
th
e
d
ataset.
T
h
is
w
as
d
o
n
e
th
r
o
u
g
h
a
s
et
o
f
r
esid
u
al
-
b
ased
d
iag
n
o
s
tics
,
in
clu
d
in
g
s
ca
tter
p
lo
ts
,
k
er
n
el
-
d
en
s
ity
cu
r
v
es,
an
d
er
r
o
r
-
d
is
tr
ib
u
tio
n
ch
ec
k
s
.
T
h
ese
an
aly
s
es
w
er
e
u
s
ed
to
d
etec
t
is
s
u
es
s
u
ch
as
h
eter
o
s
ce
d
asti
city
,
p
er
s
is
ten
t
b
ias,
o
r
u
n
u
s
u
al
er
r
o
r
p
atter
n
s
in
s
p
ec
if
ic
v
eh
icl
e
g
r
o
u
p
s
.
P
ar
ticu
lar
atten
tio
n
w
as
g
iv
en
to
s
eg
m
en
ts
th
at
o
f
ten
b
eh
av
e
d
if
f
er
en
tly
,
s
u
ch
as
h
ea
v
y
-
d
u
ty
tr
u
ck
s
,
h
y
b
r
id
v
eh
icles,
an
d
Evaluation Warning : The document was created with Spire.PDF for Python.