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
6
5
2
~
2
6
6
3
I
SS
N:
2088
-
8
7
0
8
,
DOI
: 1
0
.
1
1
5
9
1
/ijece.
v
16
i
5
.
pp
2
6
5
2
-
2
6
6
3
2652
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ij
ec
e.
ia
esco
r
e.
co
m
An arti
ficial int
ell
ig
ence
-
ba
sed s
tacking
ensemble
f
ra
mewo
rk
for sma
rt
ir
rig
a
tion pump
cont
ro
l
using
Io
T sens
o
r data
Sa
rr
a
G
o
ura
ri
1,
2
,
Wa
f
a
Dif
a
l
la
h
2,
3
,
B
elk
a
ce
m
Dr
a
o
ui
3
1
La
b
o
r
a
t
o
r
y
o
f
I
n
f
o
r
ma
t
i
o
n
P
r
o
c
e
ss
i
n
g
a
n
d
T
e
l
e
c
o
mm
u
n
i
c
a
t
i
o
n
(
L
TI
T)
,
D
e
p
a
r
t
m
e
n
t
o
f
M
a
t
h
e
ma
t
i
c
s
a
n
d
C
o
mp
u
t
e
r
S
c
i
e
n
c
e
,
F
a
c
u
l
t
y
o
f
E
x
a
c
t
S
c
i
e
n
c
e
s
,
T
a
h
r
i
M
o
h
a
mm
e
d
U
n
i
v
e
r
si
t
y
o
f
B
e
c
h
a
r
,
B
e
c
h
a
r
,
A
l
g
e
r
i
a
2
I
n
n
o
v
a
t
i
o
n
i
n
I
n
f
o
r
ma
t
i
c
s
a
n
d
E
n
g
i
n
e
e
r
i
n
g
La
b
o
r
a
t
o
r
y
(
I
N
I
E
LA
B
)
,
D
e
p
a
r
t
men
t
o
f
M
a
t
h
e
ma
t
i
c
s
a
n
d
C
o
m
p
u
t
e
r
S
c
i
e
n
c
e
,
F
a
c
u
l
t
y
o
f
E
x
a
c
t
S
c
i
e
n
c
e
s
,
T
a
h
r
i
M
o
h
a
me
d
U
n
i
v
e
r
si
t
y
,
B
e
c
h
a
r
,
A
l
g
e
r
i
a
3
L
a
b
o
r
a
t
o
r
y
o
f
E
n
e
r
g
e
t
i
c
i
n
A
r
i
d
Z
o
n
e
s (
EN
E
R
G
A
R
I
D
)
,
F
a
c
u
l
t
y
o
f
T
e
c
h
n
o
l
o
g
y
,
T
a
h
r
i
M
o
h
a
mm
e
d
U
n
i
v
e
r
s
i
t
y
o
f
B
e
c
h
a
r
,
B
e
c
h
a
r
,
A
l
g
e
r
i
a
Art
icle
I
nfo
AB
S
T
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
Ap
r
3
0
,
2
0
2
6
R
ev
is
ed
Au
g
1
3
,
2
0
2
6
Acc
ep
ted
Au
g
1
9
,
2
0
2
6
Eff
icie
n
t
wa
ter
m
a
n
a
g
e
m
e
n
t
i
s
e
ss
e
n
ti
a
l
fo
r
su
sta
in
a
b
le
a
g
ricu
lt
u
ra
l
p
ro
d
u
c
ti
o
n
,
p
a
rti
c
u
larl
y
in
a
ri
d
a
n
d
se
m
i
-
a
rid
re
g
i
o
n
s
wh
e
re
wa
ter
re
so
u
rc
e
s
a
re
li
m
it
e
d
.
M
a
c
h
in
e
-
lea
rn
i
n
g
-
b
a
se
d
irri
g
a
ti
o
n
s
y
ste
m
s
c
a
n
su
p
p
o
rt
a
u
to
m
a
ted
p
u
m
p
-
o
p
e
ra
ti
o
n
d
e
c
isio
n
s
u
si
n
g
e
n
v
ir
o
n
m
e
n
tal
a
n
d
so
il
-
re
late
d
se
n
so
r
d
a
ta.
H
o
we
v
e
r,
m
o
st
p
re
v
io
u
s
stu
d
ies
h
a
v
e
f
o
c
u
se
d
o
n
i
n
d
i
v
id
u
a
l
m
a
c
h
in
e
-
lea
rn
in
g
m
o
d
e
ls,
wh
il
e
th
e
a
p
p
l
ica
ti
o
n
o
f
sta
c
k
e
d
e
n
s
e
m
b
les
to
b
in
a
ry
irri
g
a
ti
o
n
p
u
m
p
-
sta
tu
s
p
re
d
ictio
n
re
m
a
in
s
re
lati
v
e
ly
li
m
it
e
d
.
Th
is
stu
d
y
p
r
o
p
o
se
s
a
sta
c
k
in
g
-
b
a
se
d
fra
m
e
wo
rk
fo
r
p
re
d
icti
n
g
irri
g
a
t
io
n
p
u
m
p
o
p
e
ra
ti
o
n
in
a
n
ON
/OF
F
c
las
sifica
ti
o
n
se
tt
in
g
.
T
h
e
fra
m
e
w
o
rk
u
se
s
e
n
v
iro
n
m
e
n
tal
a
n
d
so
i
l
-
re
late
d
v
a
riab
les
,
in
c
l
u
d
i
n
g
so
i
l
m
o
ist
u
re
,
tem
p
e
ra
tu
re
,
h
u
m
id
it
y
,
a
n
d
a
n
u
m
e
rica
l
ti
m
e
-
re
late
d
fe
a
tu
re
.
Ra
n
d
o
m
f
o
re
st
(RF
)
,
e
x
trem
e
g
ra
d
ien
t
b
o
o
sti
n
g
(
XG
Bo
o
st
)
,
a
n
d
m
u
lt
il
a
y
e
r
p
e
rc
e
p
tro
n
m
o
d
e
ls
we
re
train
e
d
a
s
b
a
se
lea
rn
e
rs,
a
n
d
th
e
ir
o
u
t
-
of
-
fo
ld
(OO
F
)
p
re
d
ictio
n
s
we
re
c
o
m
b
i
n
e
d
u
sin
g
a
lo
g
isti
c
-
re
g
re
ss
io
n
m
e
ta
-
lea
rn
e
r.
Th
e
m
o
d
e
ls
we
re
e
v
a
lu
a
ted
o
n
a
h
e
ld
-
o
u
t
tes
t
se
t
u
sin
g
a
c
c
u
ra
c
y
,
p
re
c
isio
n
,
re
c
a
ll
,
sp
e
c
ifi
c
it
y
,
a
n
d
F
1
-
sc
o
re
.
Th
e
i
n
d
i
v
i
d
u
a
l
m
o
d
e
ls
a
c
h
iev
e
d
a
c
c
u
ra
c
ie
s
ra
n
g
in
g
fr
o
m
9
7
.
3
4
%
to
9
9
.
9
5
%
,
with
XG
Bo
o
st
p
ro
v
id
i
n
g
th
e
b
e
st
in
d
i
v
id
u
a
l
p
e
rfo
rm
a
n
c
e
.
T
h
e
p
ro
p
o
se
d
sta
c
k
in
g
e
n
se
m
b
le
a
c
h
ie
v
e
d
9
9
.
9
7
%
a
c
c
u
ra
c
y
,
9
9
.
9
4
%
p
re
c
isio
n
,
1
0
0
.
0
0
%
re
c
a
ll
,
9
9
.
9
4
%
sp
e
c
ifi
c
it
y
,
a
n
d
a
9
9
.
9
7
%
F
1
-
sc
o
re
.
C
o
m
p
a
re
d
wit
h
XG
Bo
o
st
,
th
e
e
n
se
m
b
le
fu
rth
e
r
re
fin
e
d
p
re
d
ictiv
e
p
e
rf
o
rm
a
n
c
e
,
imp
ro
v
i
n
g
a
c
c
u
ra
c
y
b
y
0
.
0
2
p
e
rc
e
n
tag
e
p
o
i
n
ts
a
n
d
F1
-
sc
o
re
b
y
0
.
0
1
p
e
rc
e
n
tag
e
p
o
in
ts
wh
il
e
a
c
h
ie
v
in
g
c
o
m
p
lete
e
li
m
in
a
ti
o
n
o
f
fa
lse
n
e
g
a
ti
v
e
s
(
1
0
0
.
0
0
%
re
c
a
ll
).
Th
e
se
re
su
lt
s
d
e
m
o
n
stra
te
t
h
e
p
o
ten
ti
a
l
o
f
sta
c
k
e
d
e
n
se
m
b
le
lea
rn
i
n
g
to
imp
ro
v
e
b
in
a
r
y
p
u
m
p
-
o
p
e
ra
ti
o
n
p
re
d
i
c
ti
o
n
a
n
d
su
p
p
o
rt
d
a
ta
-
d
ri
v
e
n
irri
g
a
ti
o
n
m
a
n
a
g
e
m
e
n
t
in
wa
ter
-
li
m
it
e
d
e
n
v
iro
n
m
e
n
ts.
K
ey
w
o
r
d
s
:
Ma
ch
in
e
lear
n
in
g
R
an
d
o
m
f
o
r
est
Sm
ar
t ir
r
ig
atio
n
Stack
in
g
en
s
em
b
le
XGBo
o
s
t
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
:
Sar
r
a
Go
u
r
ar
i
Dep
ar
tm
en
t o
f
Ma
th
an
d
C
o
m
p
u
ter
Scien
ce
s
,
Facu
lty
o
f
E
x
a
ct
Scien
ce
s
,
T
ah
r
i M
o
h
am
e
d
Un
iv
er
s
ity
Deb
d
ab
a
R
o
ad
,
B
ec
h
ar
0
8
0
0
0
,
Alg
er
ia
E
m
ail:
g
o
u
r
ar
i.sar
r
a@
u
n
iv
-
b
e
ch
ar
.
d
z
1.
I
NT
RO
D
UCT
I
O
N
T
h
e
f
o
u
n
d
atio
n
o
f
s
u
s
tain
ab
le
d
ev
elo
p
m
e
n
t
an
d
g
lo
b
al
f
o
o
d
s
ec
u
r
ity
co
n
tin
u
es
to
b
e
a
g
r
icu
ltu
r
e.
I
t
is
a
v
ital
s
ec
to
r
t
h
at
d
ir
ec
tly
im
p
ac
ts
f
o
o
d
s
ec
u
r
ity
an
d
s
u
s
tain
ab
le
d
e
v
elo
p
m
e
n
t.
Ho
we
v
er
,
d
u
e
to
p
o
p
u
latio
n
ex
p
an
s
io
n
an
d
clim
ate
c
h
an
g
e,
in
d
u
s
tr
y
is
e
x
p
er
ien
ci
n
g
a
r
is
in
g
s
h
o
r
tag
e
o
f
wate
r
.
T
h
is
h
as
m
a
d
e
e
f
f
icien
t
ir
r
ig
atio
n
m
a
n
ag
em
e
n
t
a
cr
iti
ca
l
p
r
io
r
ity
a
n
d
a
p
r
ess
in
g
g
l
o
b
al
ch
allen
g
e
[
1
]
.
As
a
r
esu
lt,
th
er
e
is
a
g
r
o
win
g
n
ee
d
f
o
r
s
m
ar
t ir
r
ig
atio
n
s
y
s
tem
s
th
at
in
co
r
p
o
r
ate
m
o
d
e
r
n
s
e
n
s
o
r
tech
n
o
l
o
g
y
a
n
d
d
ata
-
d
r
iv
en
d
ec
is
io
n
-
m
ak
i
n
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
A
n
a
r
tifi
cia
l in
tellig
en
ce
-
b
a
s
ed
s
ta
ck
in
g
en
s
emb
le
fr
a
mewo
r
k
fo
r
s
ma
r
t i
r
r
ig
a
tio
n
…
(
S
a
r
r
a
Go
u
r
a
r
i
)
2653
tech
n
iq
u
es
[
2
]
.
R
ec
en
t
d
ev
el
o
p
m
en
ts
in
th
e
I
n
ter
n
et
o
f
T
h
in
g
s
(
I
o
T
)
h
av
e
m
ad
e
it
f
e
asib
le
f
o
r
l
o
w
-
co
s
t
s
en
s
o
r
s
to
co
n
tin
u
o
u
s
ly
m
o
n
ito
r
im
p
o
r
tan
t
en
v
ir
o
n
m
e
n
tal
p
a
r
am
eter
s
,
s
u
ch
as
tem
p
er
atu
r
e,
h
u
m
id
ity
,
an
d
s
o
il
m
o
is
tu
r
e;
th
is
,
alo
n
g
with
th
e
in
cr
ea
s
in
g
av
ailab
ilit
y
o
f
m
ac
h
in
e
lear
n
in
g
(
ML
)
tec
h
n
iq
u
e
s
,
h
as
en
ab
led
d
ata
-
d
r
iv
en
d
ec
is
io
n
-
m
a
k
in
g
in
ir
r
ig
atio
n
s
y
s
tem
s
,
wh
ich
h
as
r
esu
lted
in
q
u
an
tifia
b
le
i
n
cr
ea
s
es
in
en
er
g
y
an
d
wate
r
ef
f
icien
cy
[
3
]
.
Nu
m
er
o
u
s
d
o
m
ain
s
,
in
clu
d
i
n
g
in
tr
u
s
i
o
n
d
etec
tio
n
a
n
d
p
o
wer
co
n
s
u
m
p
tio
n
p
r
ed
ictio
n
,
an
d
m
ac
h
in
e
lear
n
i
n
g
,
h
a
v
e
b
ee
n
wid
ely
u
s
ed
to
ass
es
s
an
d
f
o
r
ec
ast
cr
o
p
wate
r
r
eq
u
ir
e
m
e
n
ts
an
d
s
ch
ed
u
lin
g
.
Ma
n
y
ac
ad
em
ics
h
av
e
tak
e
n
an
in
ter
est
in
en
s
em
b
le
lear
n
i
n
g
m
eth
o
d
o
lo
g
ies,
wh
ich
ar
e
ex
ten
s
iv
ely
u
s
ed
to
p
r
ed
ict
th
e
q
u
an
tity
o
f
wate
r
n
ee
d
ed
to
im
p
r
o
v
e
m
o
d
elin
g
ac
cu
r
ac
y
[
4
]
.
T
h
e
i
d
ea
o
f
e
n
s
em
b
le
lear
n
i
n
g
is
t
o
co
m
b
in
e
m
an
y
s
im
p
le
m
o
d
e
ls
(
wea
k
lear
n
er
s
)
i
n
to
a
n
e
w
m
o
d
el
(
s
tr
o
n
g
lea
r
n
er
)
in
o
r
d
e
r
to
im
p
r
o
v
e
p
r
ed
ictio
n
r
esu
lts
,
m
in
im
ize
d
iv
er
g
en
ce
,
o
r
d
ec
r
ea
s
e
v
a
r
ian
ce
.
T
h
e
th
r
ee
m
ain
ca
te
g
o
r
ies
o
f
en
s
em
b
le
lear
n
in
g
tech
n
iq
u
es a
r
e
s
tack
i
n
g
,
b
o
o
s
tin
g
,
a
n
d
b
a
g
g
in
g
[
5
]
.
Ma
n
y
s
tu
d
ies
ar
e
b
ein
g
co
n
d
u
cted
o
n
th
e
i
n
co
r
p
o
r
atio
n
o
f
I
o
T
a
n
d
AI
in
to
s
m
ar
t
ag
r
ic
u
ltu
r
e.
T
h
is
s
ec
tio
n
co
v
er
s
s
o
m
e
o
f
th
e
m
o
s
t
s
ig
n
if
ican
t
cu
r
r
e
n
t
s
tu
d
ies
in
th
is
ar
ea
.
T
h
e
au
t
h
o
r
s
in
[
6
]
s
u
g
g
est
a
s
y
s
tem
th
at
m
o
n
ito
r
s
tem
p
er
atu
r
e
an
d
wate
r
ev
ap
o
r
atio
n
r
ates
u
s
in
g
tem
p
e
r
atu
r
e
g
au
g
es
an
d
s
en
s
o
r
s
,
an
d
d
etec
ts
ex
ce
s
s
wate
r
.
I
n
[
7
]
,
th
e
au
th
o
r
s
u
s
ed
clo
u
d
co
m
p
u
tin
g
an
d
I
o
T
to
o
p
tim
ize
wate
r
m
an
ag
e
m
en
t
in
ag
r
icu
ltu
r
e,
in
teg
r
atin
g
a
n
E
SP
3
2
with
DHT
2
2
,
m
o
is
tu
r
e,
an
d
s
en
s
o
r
s
f
o
r
m
o
n
ito
r
in
g
wate
r
lev
e
ls
in
r
ea
l
tim
e
f
o
r
ir
r
ig
atio
n
.
B
astam
et
a
l.
[
8
]
d
ev
elo
p
e
d
a
h
ig
h
-
ac
cu
r
ac
y
ir
r
ig
atio
n
m
o
d
el
u
tili
zin
g
a
cu
ttin
g
-
ed
g
e
m
ac
h
in
e
lear
n
in
g
tech
n
i
q
u
e.
T
h
eir
h
y
b
r
id
en
s
em
b
le
ap
p
r
o
ac
h
o
u
tp
e
r
f
o
r
m
e
d
s
o
lo
m
o
d
els
to
attai
n
9
9
%
ac
cu
r
ac
y
in
d
y
n
am
ically
f
o
r
ec
asti
n
g
cr
o
p
wate
r
d
em
an
d
s
b
y
co
m
b
in
in
g
s
tack
in
g
an
d
a
d
ec
is
io
n
tr
ee
v
o
tin
g
lay
er
.
I
n
o
r
d
e
r
to
f
o
r
ec
ast
cr
o
p
wate
r
r
eq
u
ir
e
m
en
ts
,
r
eg
r
ess
io
n
-
b
ased
ir
r
ig
a
tio
n
s
y
s
tem
s
u
s
e
p
ast
d
ata
f
r
o
m
s
o
il,
p
lan
t,
an
d
clim
ate
s
en
s
o
r
s
.
T
h
e
au
th
o
r
s
also
d
is
clo
s
ed
th
at
a
th
o
r
o
u
g
h
an
aly
s
is
o
f
th
e
ass
o
ciate
d
v
ar
iab
les
is
n
ec
ess
ar
y
,
as is
th
e
cr
ea
tio
n
o
f
a
tr
u
s
two
r
th
y
m
o
d
el
f
o
r
ac
cu
r
ately
ca
lc
u
latin
g
th
e
wate
r
d
em
a
n
d
[
9
]
.
I
n
o
r
d
e
r
to
o
v
e
r
co
m
e
th
e
d
r
a
wb
ac
k
s
o
f
s
in
g
le
-
m
o
d
el
tech
n
iq
u
es
in
d
iv
er
s
e
ag
r
ic
u
ltu
r
al
co
n
tex
ts
,
r
esear
ch
er
s
h
av
e
m
o
r
e
r
ec
e
n
tly
lo
o
k
ed
at
h
y
b
r
id
an
d
en
s
e
m
b
le
ap
p
r
o
ac
h
es.
Fo
r
in
s
tan
ce
,
u
s
in
g
an
Ar
d
u
in
o
UNO
an
d
No
d
eM
C
U
to
lev
er
ag
e
E
d
g
e
C
o
m
p
u
tin
g
,
a
two
-
p
h
ase
I
o
T
-
en
ab
le
d
ir
r
ig
atio
n
s
y
s
tem
r
ed
u
ce
d
wate
r
u
s
e
b
y
2
5
%,
in
cr
ea
s
ed
cr
o
p
o
u
tp
u
t b
y
1
8
%,
an
d
im
p
r
o
v
ed
en
er
g
y
ef
f
icien
cy
b
y
3
5
%
[
1
0
]
.
E
s
m
ail
et
a
l.
[
1
1
]
d
ev
elo
p
e
d
an
au
t
o
m
ated
ir
r
ig
atio
n
s
y
s
t
em
th
at
u
s
es
m
ac
h
in
e
lear
n
in
g
an
d
I
o
T
s
en
s
o
r
s
to
en
h
an
ce
ir
r
ig
atio
n
s
ch
ed
u
lin
g
.
R
ea
l
-
tim
e
d
ata
was
g
ath
er
ed
b
y
s
en
s
o
r
s
,
s
u
ch
as
am
b
ien
t
an
d
s
o
il
m
o
is
tu
r
e
s
en
s
o
r
s
,
an
d
p
r
ep
r
o
ce
s
s
ed
b
ef
o
r
e
b
ein
g
in
p
u
t
in
to
m
ac
h
in
e
lear
n
in
g
m
o
d
els
lik
e
s
u
p
p
o
r
t
v
ec
to
r
m
ac
h
in
es
(
SVM)
,
lo
g
is
tic
r
eg
r
ess
io
n
,
an
d
d
ec
is
io
n
tr
ee
s
.
W
ith
an
ac
cu
r
ac
y
o
f
8
9
.
6
8
%
,
th
e
d
ec
is
io
n
tr
ee
m
o
d
el
f
ar
ed
b
etter
th
an
th
e
o
t
h
er
s
.
B
en
za
o
u
ia
et
a
l.
in
[
1
2
]
cr
ea
ted
a
Sm
ar
tTe
ch
-
Ag
r
i
g
a
d
g
et
th
at
u
s
es
th
e
I
o
T
to
tr
ac
k
air
tem
p
er
atu
r
e,
s
o
il
m
o
is
tu
r
e,
h
u
m
id
ity
,
s
m
o
k
e
d
e
tectio
n
,
an
d
an
im
al
b
eh
av
io
r
i
n
ag
r
icu
ltu
r
al
ar
ea
s
,
r
ed
u
cin
g
la
b
o
r
c
o
s
ts
an
d
wate
r
waste.
I
n
co
m
p
ar
is
o
n
to
th
e
s
u
g
g
ested
f
r
am
ewo
r
k
,
th
is
wo
r
k
'
s
s
ca
lab
il
ity
an
d
p
r
ec
is
io
n
ar
e
lim
ited
b
y
its
ab
s
en
ce
o
f
s
o
p
h
is
ticated
m
u
l
ti
-
s
en
s
o
r
d
ata
p
r
e
d
ictio
n
o
f
i
r
r
ig
atio
n
b
ased
o
n
m
ac
h
in
e
lear
n
in
g
,
as we
ll a
s
en
er
g
y
-
s
av
i
n
g
f
ea
t
u
r
es in
clu
d
in
g
lo
w
-
p
o
wer
m
o
d
es a
n
d
d
y
n
a
m
ic
th
r
esh
o
ld
s
.
M
a
j
u
m
d
a
r
e
t
a
l
.
[
1
3
]
p
r
e
s
e
n
te
d
a
h
y
b
r
i
d
p
r
e
d
i
c
t
i
o
n
f
r
a
m
e
w
o
r
k
t
h
a
t
i
m
p
r
o
v
e
s
t
h
e
es
t
i
m
a
t
e
o
f
s
o
i
l
m
o
i
s
t
u
r
e
a
n
d
e
v
a
p
o
t
r
a
n
s
p
i
r
a
tio
n
a
t
v
a
r
i
o
u
s
p
h
a
s
e
s
o
f
r
i
c
e
g
r
o
w
t
h
b
y
c
o
m
b
i
n
i
n
g
e
n
s
e
m
b
l
e
m
a
c
h
i
n
e
l
e
a
r
n
i
n
g
t
e
c
h
n
i
q
u
e
s
wi
t
h
a
s
a
l
p
s
wa
r
m
-
b
a
s
e
d
f
e
a
t
u
r
e
o
p
t
i
m
i
za
t
io
n
a
p
p
r
o
a
c
h
.
T
h
e
i
r
s
t
r
a
t
e
g
y
e
n
h
a
n
c
e
s
m
o
d
e
l
d
e
p
e
n
d
a
b
i
l
i
t
y
a
n
d
p
r
e
d
i
c
ti
v
e
p
e
r
f
o
r
m
a
n
c
e
b
y
f
i
n
e
-
t
u
n
i
n
g
t
h
e
s
e
le
c
t
i
o
n
o
f
i
m
p
o
r
t
an
t
a
g
r
o
n
o
m
i
c
a
n
d
e
n
v
i
r
o
n
m
e
n
t
a
l
v
a
r
i
a
b
l
es
,
d
e
m
o
n
s
t
r
a
t
i
n
g
t
h
e
p
r
o
m
i
s
e
o
f
i
n
te
g
r
a
t
i
n
g
o
p
t
i
m
i
z
a
ti
o
n
t
e
c
h
n
i
q
u
e
s
w
i
t
h
e
n
s
e
m
b
l
e
l
e
a
r
n
i
n
g
f
o
r
m
o
r
e
p
r
e
c
i
s
e
i
r
r
ig
a
t
i
o
n
-
r
e
l
a
t
e
d
f
o
r
e
c
as
t
i
n
g
.
R
u
b
o
a
n
d
Z
i
n
k
e
r
n
a
g
e
l
[
1
4
]
c
r
e
a
t
e
d
n
e
u
r
a
l
n
e
t
w
o
r
k
m
o
d
e
l
s
t
o
f
o
r
e
c
as
t
a
v
ai
l
a
b
l
e
wa
t
e
r
c
a
p
a
c
it
y
f
o
r
i
r
r
i
g
a
ti
o
n
s
c
h
ed
u
l
i
n
g
,
i
n
c
o
r
p
o
r
a
t
i
n
g
a
m
u
l
t
il
ay
e
r
p
e
r
c
e
p
t
r
o
n
a
n
d
a
l
o
n
g
s
h
o
r
t
-
t
e
r
m
m
e
m
o
r
y
.
T
h
e
i
r
m
e
t
h
o
d
,
w
h
i
c
h
w
a
s
b
a
s
e
d
o
n
e
n
s
e
m
b
l
e
m
o
d
e
l
i
n
g
a
n
d
t
r
a
n
s
f
e
r
l
e
a
r
n
i
n
g
f
r
o
m
m
u
l
t
i
-
l
o
c
a
ti
o
n
d
a
t
a
,
p
r
o
d
u
c
e
d
g
o
o
d
a
c
c
u
r
a
c
y
(
R
²
>
0
.
9
8
,
R
MS
E
<
1
.
5
%
)
a
n
d
d
e
m
o
n
s
t
r
a
te
d
t
h
e
s
i
g
n
i
f
i
c
a
n
c
e
o
f
t
e
m
p
o
r
a
l
d
y
n
a
m
i
c
s
i
n
s
o
il
m
o
i
s
t
u
r
e
p
r
e
d
i
c
t
i
o
n
.
K
a
n
n
a
n
e
t
a
l
.
[
1
5
]
s
u
g
g
e
s
t
e
d
a
s
t
a
c
k
e
d
e
n
s
em
b
l
e
m
o
d
e
l
-
b
a
s
e
d
s
m
a
r
t
w
a
t
e
r
m
a
n
a
g
e
m
e
n
t
p
a
r
ad
i
g
m
f
o
r
a
g
r
i
c
u
l
t
u
r
al
i
r
r
i
g
a
t
i
o
n
b
a
s
e
d
o
n
m
a
c
h
i
n
e
l
e
a
r
n
i
n
g
.
M
a
j
u
m
d
a
r
e
t
a
l
.
[
1
6
]
d
e
v
e
l
o
p
e
d
a
m
a
c
h
i
n
e
l
e
a
r
n
i
n
g
-
b
a
s
e
d
m
e
t
h
o
d
t
o
f
o
r
e
c
a
s
t
r
i
c
e'
s
r
e
q
u
i
r
e
m
e
n
t
s
f
o
r
f
e
r
t
i
l
i
z
e
r
a
n
d
i
r
r
i
g
a
t
i
o
n
w
a
t
e
r
a
t
d
i
f
f
e
r
e
n
t
s
t
a
g
es
o
f
d
e
v
e
l
o
p
m
e
n
t
.
T
h
e
i
r
a
p
p
r
o
a
c
h
e
n
h
a
n
c
e
s
p
r
e
d
i
c
t
i
o
n
a
c
c
u
r
a
c
y
a
n
d
f
a
c
i
l
it
a
t
es
i
m
p
r
o
v
e
d
a
g
r
i
c
u
l
t
u
r
a
l
i
n
p
u
t
m
a
n
a
g
e
m
e
n
t
b
y
f
u
s
i
n
g
r
a
n
d
o
m
f
o
r
e
s
t
(
R
F
)
w
i
t
h
a
B
a
y
e
s
i
a
n
g
e
n
e
ti
c
o
p
t
im
i
z
a
t
i
o
n
t
e
c
h
n
i
q
u
e
.
T
h
i
s
m
i
g
h
t
l
ea
d
t
o
i
n
c
r
e
as
e
d
p
r
o
d
u
c
t
i
o
n
a
n
d
m
o
r
e
e
f
f
e
c
t
i
v
e
u
s
e
o
f
r
e
s
o
u
r
c
e
s
.
I
n
y
a
n
g
e
t
a
l
.
[
1
7
]
e
x
a
m
i
n
e
d
t
h
e
k
-
n
e
a
r
e
s
t
n
e
i
g
h
b
o
r
s
(
K
NN
)
,
SV
M
,
e
x
t
r
e
m
e
g
r
a
d
i
e
n
t
b
o
o
s
ti
n
g
(
X
G
B
o
o
s
t
)
,
a
n
d
RF
m
a
c
h
i
n
e
l
e
a
r
n
i
n
g
m
o
d
e
l
s
f
o
r
t
h
e
c
l
as
s
i
f
i
c
a
ti
o
n
o
f
i
r
r
i
g
a
t
i
o
n
m
e
t
h
o
d
s
(
o
v
e
r
h
e
a
d
,
s
u
r
f
a
c
e
,
an
d
p
r
e
c
i
s
i
o
n
)
u
s
i
n
g
a
n
a
g
r
i
c
u
l
tu
r
a
l
d
a
t
a
s
et
o
b
t
a
i
n
e
d
f
r
o
m
K
a
g
g
l
e
.
W
it
h
a
n
a
c
c
u
r
ac
y
o
f
8
6
%
a
n
d
a
K
a
p
p
a
s
c
o
r
e
o
f
0
.
7
9
,
t
h
e
i
r
f
i
n
d
i
n
g
s
s
h
o
w
e
d
t
h
a
t
e
n
s
e
m
b
l
e
a
p
p
r
o
a
c
h
e
s
,
i
n
p
a
r
ti
c
u
l
a
r
,
XGB
o
o
s
t
,
p
e
r
f
o
r
m
e
d
b
e
t
t
e
r
t
h
a
n
s
i
n
g
l
e
t
e
c
h
n
i
q
u
es
,
s
u
cc
e
s
s
f
u
l
l
y
m
a
n
a
g
i
n
g
t
h
e
d
a
t
a'
s
non
-
l
i
n
e
a
r
c
o
r
r
e
l
a
ti
o
n
s
m
o
s
t
p
r
e
v
i
o
u
s
s
t
u
d
i
e
s
h
a
v
e
f
o
c
u
s
e
d
o
n
s
i
n
g
l
e
m
o
d
e
l
s
o
r
t
r
a
d
it
i
o
n
a
l
e
n
s
e
m
b
l
e
m
e
t
h
o
d
s
,
a
n
d
e
v
e
n
t
h
e
s
t
a
c
k
i
n
g
-
b
a
s
e
d
s
tu
d
i
e
s
a
m
o
n
g
t
h
e
m
[
8
]
,
[
1
3
]
,
[
1
5
]
g
e
n
e
r
a
l
l
y
s
u
f
f
e
r
f
r
o
m
tw
o
im
p
o
r
t
a
n
t
l
i
m
it
a
t
i
o
n
s
.
F
i
r
s
t
,
t
h
e
y
e
v
al
u
a
t
e
t
h
e
c
o
n
t
r
i
b
u
t
i
o
n
o
f
i
n
d
i
v
i
d
u
a
l
b
as
e
l
ea
r
n
er
s
i
n
is
o
l
at
i
o
n
,
w
it
h
o
u
t
q
u
a
n
t
i
f
y
i
n
g
h
o
w
m
u
c
h
t
h
e
b
a
s
e
l
e
a
r
n
e
r
s
a
ct
u
a
l
l
y
d
is
a
g
r
e
e
w
i
t
h
o
n
e
a
n
o
t
h
e
r
o
n
t
h
e
te
s
t
s
e
t
—
a
n
d
t
h
e
r
e
f
o
r
e
w
h
e
t
h
e
r
t
h
e
r
e
p
o
r
t
e
d
a
c
c
u
r
a
c
y
g
a
i
n
r
e
f
l
e
c
ts
g
e
n
u
i
n
e
c
o
m
p
le
m
e
n
t
a
r
y
i
n
f
o
r
m
a
t
i
o
n
o
r
i
s
s
i
m
p
l
y
a
n
a
v
e
r
a
g
i
n
g
e
f
f
e
ct
o
v
e
r
n
e
a
r
-
i
d
e
n
t
i
c
a
l
p
r
e
d
i
c
t
i
o
n
s
.
Se
c
o
n
d
,
d
e
p
l
o
y
m
e
n
t
f
e
a
s
i
b
il
i
t
y
is
r
a
r
e
l
y
a
d
d
r
e
s
s
e
d
q
u
a
n
t
i
t
a
ti
v
e
l
y
:
i
n
f
e
r
e
n
ce
l
a
t
e
n
c
y
,
m
e
m
o
r
y
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
6
5
2
-
2
6
6
3
2654
f
o
o
t
p
r
i
n
t
,
a
n
d
m
o
d
e
l
c
o
m
p
l
e
x
i
t
y
a
r
e
s
e
l
d
o
m
r
e
p
o
r
t
e
d
,
e
v
e
n
t
h
o
u
g
h
t
h
e
s
e
f
a
c
t
o
r
s
a
r
e
e
s
s
e
n
tia
l
f
o
r
j
u
s
t
i
f
y
i
n
g
a
n
e
n
s
e
m
b
l
e'
s
s
u
i
t
a
b
il
i
t
y
f
o
r
r
e
a
l
-
t
i
m
e
I
o
T
-
b
a
s
e
d
i
r
r
i
g
a
t
i
o
n
c
o
n
tr
o
l
o
n
r
e
s
o
u
r
c
e
-
c
o
n
s
t
r
a
i
n
e
d
e
d
g
e
h
a
r
d
w
a
r
e
.
T
h
is
o
b
s
e
r
v
a
t
i
o
n
h
i
g
h
l
i
g
h
t
s
a
g
a
p
i
n
t
h
e
r
e
s
e
a
r
c
h
l
i
te
r
a
t
u
r
e
,
w
h
i
c
h
t
h
e
p
r
e
s
e
n
t
s
t
u
d
y
a
d
d
r
e
s
s
es
t
h
r
o
u
g
h
t
h
e
c
o
n
t
r
i
b
u
t
i
o
n
s
s
u
m
m
a
r
i
z
e
d
.
Giv
en
th
e
ef
f
ec
tiv
en
ess
o
f
th
e
s
e
en
s
em
b
le
lear
n
i
n
g
s
tr
ateg
ie
s
in
im
p
r
o
v
in
g
p
r
ed
ictiv
e
p
er
f
o
r
m
an
ce
,
th
is
s
tu
d
y
b
u
ild
s
u
p
o
n
th
ese
co
n
ce
p
ts
to
ad
d
r
ess
ir
r
ig
atio
n
m
an
ag
em
en
t
c
h
allen
g
es.
T
h
e
p
r
im
ar
y
o
b
jectiv
e
o
f
th
is
p
r
o
ject
is
to
cr
ea
te
an
i
n
tellig
en
t
ir
r
ig
atio
n
d
ec
is
io
n
-
s
u
p
p
o
r
t
s
y
s
tem
th
at
ca
n
p
r
e
d
ict
p
u
m
p
o
p
er
atio
n
(
ON/OFF)
b
ased
o
n
en
v
ir
o
n
m
en
tal
an
d
s
o
il sen
s
o
r
d
ata.
T
h
e
p
r
ed
ictio
n
s
f
r
o
m
RF
,
XGBo
o
s
t,
an
d
Mu
lti
-
L
ay
er
Per
ce
p
tr
o
n
s
ar
e
m
e
r
g
ed
an
d
u
s
ed
as
a
m
eta
-
lear
n
er
'
s
in
p
u
t
ch
ar
ac
ter
is
tics
(
L
o
g
is
tic
R
eg
r
ess
io
n
)
as
p
ar
t
o
f
a
s
tack
in
g
en
s
em
b
le
m
eth
o
d
.
I
n
s
t
e
a
d
o
f
d
e
p
e
n
d
i
n
g
o
n
a
s
i
n
g
l
e
m
o
d
el
,
t
h
i
s
m
e
t
h
o
d
m
a
k
e
s
u
s
e
o
f
t
h
e
c
o
m
p
l
e
m
e
n
t
a
r
y
c
a
p
a
b
il
i
ti
e
s
o
f
b
a
s
e
l
e
a
r
n
e
r
s
t
o
i
n
c
r
e
as
e
th
e
g
e
n
e
r
a
l
i
z
at
i
o
n
a
n
d
r
o
b
u
s
t
n
e
s
s
o
f
p
u
m
p
s
t
a
t
u
s
p
r
e
d
i
c
t
i
o
n
.
I
n
a
d
d
i
t
i
o
n
,
f
e
a
t
u
r
e
i
m
p
o
r
t
a
n
c
e
a
n
a
l
y
s
i
s
a
n
d
f
e
a
t
u
r
e
s
e
l
e
ct
i
o
n
m
e
t
h
o
d
s
a
r
e
e
m
p
l
o
y
e
d
t
o
h
i
g
h
l
i
g
h
t
t
h
e
m
o
s
t
i
n
f
l
u
e
n
ti
a
l
v
a
r
i
a
b
l
e
s
a
n
d
im
p
r
o
v
e
i
n
t
e
r
p
r
e
t
a
b
i
l
it
y
.
T
h
e
m
a
i
n
c
o
n
t
r
i
b
u
t
i
o
n
s
o
f
t
h
i
s
s
t
u
d
y
a
r
e
s
u
m
m
a
r
i
z
e
d
as
:
a.
A
s
tack
in
g
en
s
em
b
le
co
m
b
in
i
n
g
RF
,
XGBo
o
s
t,
an
d
a
m
u
ltil
ay
er
p
er
ce
p
tr
o
n
is
p
r
o
p
o
s
ed
a
n
d
ev
alu
ated
f
o
r
b
in
ar
y
ir
r
ig
atio
n
p
u
m
p
-
s
tatu
s
(
ON/OFF)
p
r
ed
ictio
n
,
a
t
ask
f
o
r
wh
ic
h
s
tack
ed
e
n
s
em
b
les
r
em
ain
co
m
p
ar
ativ
ely
u
n
d
e
r
-
ex
p
lo
r
ed
r
elativ
e
to
s
in
g
le
-
m
o
d
el
ap
p
r
o
ac
h
es.
b.
A
d
ep
lo
y
m
en
t
-
o
r
ien
ted
ev
alu
atio
n
is
p
r
o
v
id
ed
,
r
ep
o
r
tin
g
p
er
-
m
o
d
el
in
f
er
en
ce
laten
cy
,
th
r
o
u
g
h
p
u
t,
an
d
s
er
ialized
m
em
o
r
y
f
o
o
tp
r
in
t,
t
o
ass
ess
f
ea
s
ib
ilit
y
f
o
r
I
o
T
/ed
g
e
d
ep
lo
y
m
en
t
r
ath
e
r
th
an
r
e
p
o
r
tin
g
p
r
e
d
ictiv
e
ac
cu
r
ac
y
alo
n
e.
c.
T
h
e
p
er
f
o
r
m
an
ce
g
ain
o
f
th
e
s
tack
in
g
en
s
em
b
le
is
ex
p
l
ain
ed
m
ec
h
an
is
tically
th
r
o
u
g
h
a
b
ase
-
lear
n
er
p
r
ed
ictio
n
-
a
g
r
ee
m
en
t
an
al
y
s
is
,
q
u
an
tify
in
g
h
o
w
m
u
ch
c
o
m
p
lem
en
tar
y
in
f
o
r
m
atio
n
t
h
e
m
eta
-
lear
n
er
ex
p
lo
its
,
r
ath
er
th
a
n
r
e
p
o
r
tin
g
th
e
ag
g
r
e
g
ate
ac
cu
r
ac
y
im
p
r
o
v
em
en
t a
lo
n
e
.
d.
An
ex
p
an
d
e
d
Sh
ap
ley
ad
d
it
iv
e
ex
p
lan
atio
n
s
(
SHAP
)
-
b
a
s
ed
in
ter
p
r
etab
ilit
y
an
aly
s
is
is
co
n
d
u
cted
,
co
m
b
in
in
g
g
lo
b
al
f
ea
tu
r
e
im
p
o
r
tan
ce
,
d
ir
ec
tio
n
-
of
-
ef
f
e
ct
(
b
ee
s
war
m
)
v
is
u
aliza
tio
n
,
an
d
f
ea
tu
r
e
-
d
ep
en
d
e
n
ce
p
lo
ts
,
lin
k
i
n
g
m
o
d
el
b
eh
av
io
u
r
to
p
h
y
s
ically
p
la
u
s
ib
le
ir
r
ig
atio
n
tr
ig
g
er
s
.
T
h
e
s
u
g
g
ested
tech
n
iq
u
e,
in
c
lu
d
in
g
d
ata
p
r
etr
ea
tm
en
t,
m
o
d
el
b
u
ild
in
g
,
a
n
d
ass
ess
m
en
t
cr
iter
ia,
is
ex
p
lain
ed
i
n
s
ec
tio
n
2
.
T
h
e
o
u
tco
m
es
o
f
th
e
e
x
p
er
im
e
n
t
ar
e
r
ep
o
r
ted
a
n
d
d
is
cu
s
s
ed
in
s
ec
tio
n
3
.
T
h
e
p
ap
er
'
s
k
ey
co
n
clu
s
io
n
s
ar
e
s
u
m
m
ar
iz
ed
in
s
ec
tio
n
4
.
2.
M
E
T
H
O
D
T
h
e
p
r
o
p
o
s
ed
a
r
ch
itectu
r
e
wa
s
d
esig
n
ed
to
s
u
p
p
o
r
t
in
tellig
e
n
t
ir
r
ig
atio
n
c
o
n
tr
o
l
b
y
p
r
ed
ic
tin
g
p
u
m
p
o
p
er
atio
n
s
tates
(
ON/OFF)
u
s
in
g
s
en
s
o
r
-
b
ased
en
v
ir
o
n
m
en
t
al
d
ata
,
as
illu
s
tr
ated
in
Fig
u
r
e
1
.
T
h
e
wo
r
k
f
lo
w
b
eg
in
s
with
d
ata
p
r
ep
r
o
ce
s
s
in
g
,
wh
er
e
r
aw
m
ea
s
u
r
em
e
n
ts
s
u
ch
as
s
o
il
m
o
is
tu
r
e,
te
m
p
er
a
tu
r
e,
an
d
h
u
m
id
ity
ar
e
p
r
ep
ar
e
d
b
ef
o
r
e
m
o
d
el
tr
a
in
in
g
.
T
h
r
ee
d
is
tin
ct
m
ac
h
in
e
lear
n
in
g
m
o
d
els
—
RF
,
XGBo
o
s
t,
an
d
m
u
ltil
ay
er
p
er
ce
p
tr
o
n
(
MLP
)
—
wer
e
em
p
lo
y
ed
as
b
ase
lear
n
er
s
d
u
e
to
th
eir
ab
ilit
y
to
ca
p
tu
r
e
i
n
tr
icate
,
n
o
n
-
lin
ea
r
p
atter
n
s
in
ir
r
ig
atio
n
d
ata.
T
h
e
d
ataset
was
th
en
s
p
lit
in
to
8
0
%
f
o
r
tr
ai
n
in
g
a
n
d
2
0
%
f
o
r
t
esti
n
g
.
T
o
im
p
r
o
v
e
p
r
ed
ictio
n
p
e
r
f
o
r
m
an
ce
an
d
r
ed
u
ce
o
v
e
r
f
itti
n
g
,
o
u
t
-
of
-
f
o
l
d
(
OOF
)
s
tack
in
g
was
ap
p
lied
to
co
m
b
in
e
th
e
o
u
tp
u
ts
o
f
th
e
b
ase
m
o
d
els.
T
h
e
g
en
e
r
ated
p
r
ed
ictio
n
s
w
er
e
s
u
b
s
eq
u
en
tly
u
s
ed
as
in
p
u
ts
f
o
r
a
L
o
g
is
tic
R
eg
r
ess
io
n
m
o
d
el
ac
tin
g
as
t
h
e
m
eta
-
lear
n
e
r
.
Fin
ally
,
th
e
s
u
g
g
ested
f
r
am
ew
o
r
k
was
ass
ess
ed
u
s
in
g
s
ev
er
al
p
er
f
o
r
m
an
ce
cr
iter
ia,
i
n
clu
d
in
g
F1
-
s
co
r
e
,
ac
cu
r
ac
y
,
p
r
ec
is
io
n
,
r
ec
all,
r
ec
eiv
e
r
o
p
er
atin
g
c
h
ar
ac
ter
is
tic
(
R
OC
)
an
aly
s
is
,
an
d
co
n
f
u
s
io
n
m
atr
i
x
ev
alu
atio
n
,
t
o
d
eter
m
i
n
e
th
e
f
i
n
al
p
u
m
p
ON/OFF p
r
ed
ictio
n
p
er
f
o
r
m
an
ce
.
Fig
u
r
e
1
.
A
r
c
h
itectu
r
e
o
f
th
e
p
r
o
p
o
s
ed
s
tack
in
g
-
b
ased
e
n
s
em
b
le
f
r
am
ewo
r
k
f
o
r
s
m
ar
t ir
r
ig
a
tio
n
p
u
m
p
co
n
tr
o
l
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
A
n
a
r
tifi
cia
l in
tellig
en
ce
-
b
a
s
ed
s
ta
ck
in
g
en
s
emb
le
fr
a
mewo
r
k
fo
r
s
ma
r
t i
r
r
ig
a
tio
n
…
(
S
a
r
r
a
Go
u
r
a
r
i
)
2655
2
.
1
.
Da
t
a
s
et
des
cr
iptio
n a
nd
pre
-
pro
ce
s
s
ing
T
h
e
d
a
tase
t
u
s
e
d
i
n
t
h
is
s
t
u
d
y
was
o
b
t
ai
n
e
d
f
r
o
m
t
h
e
p
u
b
lis
h
e
d
wo
r
k
in
[
1
8
]
,
b
ase
d
o
n
th
e
p
u
b
l
icl
y
av
ail
ab
le
s
m
a
r
t
i
r
r
ig
ati
o
n
d
at
aset
o
r
ig
in
all
y
i
n
t
r
o
d
u
ce
d
b
y
M
itta
l
[
1
9
]
a
n
d
s
u
p
p
le
m
e
n
t
ed
w
it
h
d
ata
f
r
o
m
Kag
g
l
e.
I
t
co
m
p
r
is
es
a
p
p
r
o
x
i
m
at
el
y
9
9
,
7
9
9
s
a
m
p
les
(
a
f
t
er
r
em
o
v
i
n
g
2
0
1
d
u
p
lic
at
e
r
ec
o
r
d
s
)
c
o
ll
ec
t
e
d
f
r
o
m
a
n
I
o
T
-
b
ase
d
s
m
a
r
t
ir
r
i
g
a
ti
o
n
s
y
s
t
em
,
e
ac
h
d
es
cr
ib
e
d
b
y
f
o
u
r
s
e
n
s
o
r
-
d
e
r
i
v
e
d
att
r
i
b
u
tes
: s
o
il
m
o
i
s
tu
r
e
,
t
em
p
e
r
at
u
r
e,
h
u
m
i
d
i
ty
,
a
n
d
a
n
u
m
er
ica
l
tim
e
a
tt
r
i
b
u
te
.
T
h
e
ti
m
e
at
tr
ib
u
t
e
r
e
p
r
ese
n
ts
a
d
is
c
r
e
te
l
o
g
g
in
g
i
n
d
e
x
r
a
th
er
t
h
an
a
ch
r
o
n
o
l
o
g
i
ca
l
ti
m
es
ta
m
p
;
ac
c
o
r
d
i
n
g
l
y
,
i
t
w
as
e
n
c
o
d
e
d
u
s
i
n
g
s
i
n
e
-
co
s
in
e
(
c
y
cli
ca
l
)
tr
a
n
s
f
o
r
m
at
io
n
i
n
t
o
tw
o
d
e
r
i
v
e
d
f
ea
t
u
r
es
,
d
en
o
t
e
d
_
a
n
d
_
,
t
o
p
r
ese
r
v
e
its
p
e
r
i
o
d
ic
n
atu
r
e;
th
ese
tw
o
v
a
r
ia
b
l
es
ar
e
r
e
f
e
r
e
n
c
e
d
d
ir
ec
t
ly
b
y
n
a
m
e
i
n
t
h
e
S
HAP
i
n
t
er
p
r
et
a
b
ili
ty
a
n
al
y
s
is
(
s
ec
ti
o
n
3
.
3
)
.
T
h
e
d
a
ta
s
et
was
tr
ea
t
e
d
as
a
s
tan
d
ar
d
t
ab
u
l
ar
class
if
ica
ti
o
n
p
r
o
b
l
em
r
at
h
er
t
h
an
a
ti
m
e
s
e
r
ies.
T
h
e
d
at
aset
c
o
n
t
ai
n
s
n
o
m
is
s
in
g
v
al
u
es
.
T
h
e
tar
g
et
v
ar
ia
b
le
,
p
u
m
p
s
t
at
u
s
,
is
m
o
d
e
r
a
tel
y
b
ala
n
ce
d
,
wi
t
h
5
3
,
5
4
6
ON
(
p
u
m
p
a
cti
v
e
,
5
3
.
6
5
%)
an
d
4
6
,
2
5
3
OFF
(
p
u
m
p
i
n
a
cti
v
e
,
4
6
.
3
5
%
)
s
a
m
p
les
.
T
h
e
d
atas
et
was
s
p
l
it
i
n
t
o
t
r
a
in
in
g
(
8
0
%)
a
n
d
h
el
d
-
o
u
t
test
(
2
0
%)
s
u
b
s
ets
u
s
i
n
g
s
t
r
at
if
i
e
d
s
a
m
p
li
n
g
t
o
p
r
ese
r
v
e
c
lass
b
a
la
n
c
e
i
n
b
o
t
h
p
ar
t
iti
o
n
s
.
F
ea
tu
r
e
s
ta
n
d
a
r
d
iz
ati
o
n
(
ze
r
o
m
ea
n
,
u
n
i
t
v
a
r
i
an
ce
)
was
a
p
p
li
ed
o
n
ly
wit
h
i
n
t
h
e
m
u
lti
la
y
e
r
p
er
ce
p
t
r
o
n
p
i
p
eli
n
e,
f
itt
ed
e
x
cl
u
s
i
v
e
ly
o
n
t
h
e
t
r
ai
n
i
n
g
p
a
r
ti
ti
o
n
t
o
p
r
ev
e
n
t d
ata
l
ea
k
a
g
e;
t
h
e
t
r
ee
-
b
ase
d
m
o
d
els
(
RF
,
XGB
o
o
s
t)
w
er
e
tr
ai
n
e
d
o
n
t
h
e
r
aw
f
ea
t
u
r
e
v
al
u
es,
co
n
s
is
t
e
n
t
wit
h
s
ta
n
d
a
r
d
p
r
ac
ti
ce
f
o
r
t
r
e
e
e
n
s
e
m
b
les
.
2
.
2
.
H
y
perpa
ra
m
e
t
er
t
un
ing
s
t
ra
t
eg
y
Fo
r
ea
c
h
b
ase
l
ea
r
n
e
r
,
h
y
p
er
p
ar
am
ete
r
s
we
r
e
s
e
le
cte
d
u
s
in
g
a
5
-
f
o
l
d
Gr
i
d
S
ea
r
c
h
C
V
(
s
c
o
r
ed
o
n
F
1
)
p
e
r
f
o
r
m
e
d
e
x
cl
u
s
i
v
e
ly
o
n
t
h
e
tr
a
in
in
g
p
a
r
tit
io
n
(
8
0
%
o
f
t
h
e
d
at
ase
t)
;
t
h
e
h
e
ld
-
o
u
t
test
s
e
t
was
n
o
t
u
s
e
d
at
a
n
y
p
o
i
n
t
d
u
r
i
n
g
m
o
d
el
s
ele
cti
o
n
,
p
r
e
v
e
n
ti
n
g
o
p
t
im
is
ti
c
b
i
as.
T
a
b
le
1
r
e
p
o
r
ts
t
h
e
s
e
lec
te
d
c
o
n
f
ig
u
r
ati
o
n
f
o
r
e
v
e
r
y
b
as
e
l
ea
r
n
e
r
to
g
et
h
e
r
w
it
h
th
e
m
et
a
-
lea
r
n
e
r
u
s
ed
b
y
t
h
e
St
ac
k
in
g
e
n
s
e
m
b
le
.
T
a
b
le
1
.
Se
lec
te
d
h
y
p
e
r
p
a
r
a
m
e
ter
s
f
o
r
t
h
e
b
ase
le
a
r
n
er
s
a
n
d
t
h
e
s
ta
c
k
i
n
g
m
et
a
-
lea
r
n
e
r
M
o
d
e
l
S
e
l
e
c
t
e
d
h
y
p
e
r
p
a
r
a
m
e
t
e
r
s
5
-
f
o
l
d
C
V
F
1
(
t
r
a
i
n
s
p
l
i
t
)
RF
_
=
1
0
0
;
_
ℎ
=
2
0
;
_
_
=1
0
.
9
9
8
3
X
G
B
o
o
st
_
=
2
0
0
;
_
ℎ
=
6
;
_
=
0
.
1
0
.
9
9
9
5
M
LP
ℎ
_
_
=
(
6
4
,
3
2
)
;
a
l
p
h
a
=
0
.
0
0
0
1
0
.
9
7
2
9
S
t
a
c
k
i
n
g
(
me
t
a
-
l
e
a
r
n
e
r
)
L
o
g
i
s
t
i
c
R
e
g
r
e
ss
i
o
n
(
_
=
5
0
0
0
)
,
f
i
t
o
n
5
-
f
o
l
d
O
O
F
b
a
s
e
-
l
e
a
r
n
e
r
p
r
e
d
i
c
t
i
o
n
s
_
2
.
3
.
Ra
nd
o
m
f
o
re
s
t
(
RF
)
T
h
e
RF
en
s
em
b
l
e
m
ac
h
i
n
e
l
e
ar
n
i
n
g
te
ch
n
i
q
u
e
is
u
s
e
d
t
o
b
u
il
d
m
a
n
y
d
e
cisi
o
n
t
r
ee
s
d
u
r
i
n
g
tr
ai
n
i
n
g
.
T
h
e
p
r
o
c
ess
t
h
en
p
r
o
d
u
ce
s
t
h
e
m
o
d
e
o
f
t
h
e
class
es
(
cl
ass
i
f
ic
ati
o
n
)
o
r
t
h
e
m
ea
n
f
o
r
e
ca
s
t
(
r
e
g
r
ess
i
o
n
)
f
o
r
e
v
e
r
y
s
in
g
le
t
r
e
e.
Usi
n
g
f
ea
t
u
r
e
s
el
ec
ti
o
n
a
n
d
b
o
o
ts
tr
a
p
s
am
p
li
n
g
f
o
r
e
v
er
y
s
p
lit
,
it
a
d
d
s
u
n
p
r
e
d
i
cta
b
il
it
y
,
w
h
i
ch
less
e
n
s
o
v
e
r
f
itt
in
g
a
n
d
en
h
an
c
es
g
e
n
e
r
a
liz
ati
o
n
[
2
0
]
.
2
.
4
.
E
x
t
re
m
e
g
r
a
dient
bo
o
s
t
ing
(
XG
B
o
o
s
t
)
XGBo
o
s
t
is
o
n
e
o
f
th
e
m
o
d
e
ls
th
at
ef
f
ec
tiv
ely
u
tili
ze
s
g
r
a
d
ien
t
b
o
o
s
tin
g
m
ac
h
i
n
es.
T
h
i
s
m
o
d
el
is
b
ased
o
n
th
e
b
o
o
s
tin
g
p
r
i
n
cip
le,
wh
ich
b
u
ild
s
an
en
s
em
b
le
o
f
d
ec
is
io
n
tr
ee
s
o
n
e
af
ter
th
e
o
th
er
wh
ile
ea
ch
m
o
d
el
attem
p
ts
to
co
r
r
ec
t
th
e
er
r
o
r
s
o
f
its
p
r
ed
ec
ess
o
r
.
T
h
is
m
o
d
el
is
o
f
ten
ch
ar
ac
ter
ized
b
y
its
p
er
f
o
r
m
a
n
ce
,
s
ca
lab
ilit
y
,
an
d
s
p
ee
d
,
as
well
as
its
ab
ilit
y
to
ex
t
r
a
ct
in
s
ig
h
ts
f
r
o
m
p
r
ev
io
u
s
e
r
r
o
r
s
an
d
o
p
tim
ize
h
y
p
er
p
ar
am
eter
s
[
2
1
]
.
T
h
e
X
GB
o
o
s
t m
o
d
el
iter
ativ
ely
b
u
ild
s
an
en
s
em
b
le
o
f
d
ec
is
io
n
tr
ee
s
,
wh
er
e
th
e
f
in
a
l
p
r
ed
ictio
n
f
o
r
a
g
iv
e
n
s
am
p
le
ⅰ
is
th
e
s
u
m
o
f
th
e
p
r
ed
ictio
n
s
f
r
o
m
all
tr
ee
s
:
=
∑
(
)
,
∈
=
1
(
1
)
2
.
5
.
M
ultila
y
er
perc
ept
ro
n (
M
L
P
)
An
y
c
o
n
ti
n
u
o
u
s
f
u
n
cti
o
n
m
ay
b
e
ap
p
r
o
x
im
ate
d
b
y
a
n
ML
P
,
a
s
tr
ai
g
h
tf
o
r
w
ar
d
f
ee
d
f
o
r
w
a
r
d
ANN.
A
n
ML
P
is
m
a
d
e
u
p
o
f
a
n
in
p
u
t
la
y
e
r
,
a
n
o
u
t
p
u
t
l
a
y
e
r
,
a
n
d
a
t
l
ea
s
t
o
n
e
h
i
d
d
e
n
l
ay
er
.
T
h
e
in
p
u
t la
y
e
r
is
m
a
d
e
u
p
o
f
th
e
n
o
d
es
th
at
r
ec
ei
v
e
t
h
e
i
n
p
u
t
i
n
f
o
r
m
ati
o
n
.
E
v
er
y
h
i
d
d
en
lay
e
r
n
o
d
e
is
f
o
ll
o
w
ed
b
y
a
n
o
n
-
l
in
ea
r
ac
ti
v
at
io
n
f
u
n
cti
o
n
,
u
s
in
g
a
w
ei
g
h
te
d
li
n
e
ar
s
u
m
t
o
p
r
o
ce
s
s
t
h
e
v
al
u
es f
r
o
m
t
h
e
p
r
e
ce
d
i
n
g
l
ay
e
r
.
Af
te
r
b
ei
n
g
p
r
o
ce
s
s
ed
b
y
th
e
l
ast
h
id
d
e
n
la
y
er
,
th
e
o
u
tp
u
t
la
y
e
r
c
o
n
v
er
ts
th
e
p
r
o
ce
s
s
ed
d
at
a
i
n
t
o
t
h
e
d
esi
r
e
d
v
a
lu
es
.
Uti
liz
in
g
t
h
e
b
a
ck
p
r
o
p
ag
ati
o
n
a
p
p
r
o
a
c
h
,
t
h
e
al
g
o
r
it
h
m
is
t
r
a
in
e
d
[
2
2
]
.
Ou
r
M
L
P
h
as
o
n
e
h
i
d
d
e
n
la
y
e
r
.
I
n
p
u
t
x
is
m
u
lt
ip
lie
d
b
y
wei
g
h
ts
W
1
,
t
h
e
n
p
ass
ed
t
h
r
o
u
g
h
a
n
ac
ti
v
at
io
n
f
u
n
cti
o
n
σ
1
t
o
g
et
h
i
d
d
e
n
o
u
tp
u
ts
h
.
T
h
en
a
n
o
th
er
li
n
e
ar
m
a
p
p
i
n
g
a
n
d
a
s
ec
o
n
d
a
cti
v
ati
o
n
σ2
g
iv
e
th
e
f
i
n
al
p
r
e
d
i
cti
o
n
y^
:
ℎ
=
1
(
1
+
1
)
(
2
)
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
6
5
2
-
2
6
6
3
2656
^
=
2
(
2
ℎ
+
2
)
(3
)
2
.
6
.
St
a
ck
ing
e
ns
em
ble (
O
O
F
)
Stac
k
i
n
g
is
a
n
e
n
s
e
m
b
le
m
ac
h
i
n
e
lea
r
n
i
n
g
p
r
o
c
ess
t
h
a
t
u
s
es
a
m
et
ac
lass
i
f
i
er
to
m
er
g
e
m
a
n
y
r
e
g
r
ess
i
o
n
o
r
class
i
f
i
ca
ti
o
n
m
o
d
els.
T
o
b
ea
t
s
ta
n
d
al
o
n
e
m
o
d
els
i
n
m
a
k
i
n
g
p
r
e
d
i
cti
o
n
s
,
s
ta
ck
in
g
ca
n
t
ak
e
ad
v
a
n
t
a
g
e
o
f
t
h
e
s
t
r
en
g
t
h
s
o
f
m
u
lti
p
l
e
h
ig
h
l
y
ef
f
ec
ti
v
e
m
o
d
els
o
n
a
r
e
g
r
ess
i
o
n
tas
k
.
T
h
e
d
is
t
in
ct
r
e
g
r
ess
i
o
n
m
o
d
els
a
r
e
b
ase
d
o
n
th
e
o
u
tp
u
ts
(
m
et
a
-
f
e
at
u
r
es
)
o
f
t
h
e
s
e
p
a
r
ate
m
o
d
els
,
a
f
te
r
w
h
i
ch
a
m
et
a
class
i
f
i
er
is
a
p
p
li
ed
b
as
ed
o
n
t
h
e
f
u
ll
tr
ai
n
i
n
g
d
at
aset
[
2
3
]
.
T
o
f
u
r
t
h
e
r
im
p
r
o
v
e
p
r
e
d
i
cti
o
n
a
cc
u
r
a
cy
,
t
h
e
t
h
r
e
e
b
ase
cl
ass
i
f
ie
r
s
—
XGBo
o
s
t
,
RF
,
a
n
d
t
h
e
m
u
lti
la
y
e
r
p
er
ce
p
t
r
o
n
—
we
r
e
c
o
m
b
in
ed
i
n
t
o
a
n
OOF
s
ta
c
k
i
n
g
e
n
s
e
m
b
le
.
T
h
e
t
r
ai
n
i
n
g
d
at
a
i
n
t
h
is
s
t
u
d
y
was
f
i
r
s
t
s
u
b
j
ec
t
ed
t
o
K
-
f
o
ld
c
r
o
s
s
-
v
al
id
ati
o
n
,
m
a
k
i
n
g
c
er
tai
n
t
h
a
t
t
h
e
p
u
m
p
o
n
/
o
f
f
la
b
e
l
cl
ass
d
is
t
r
i
b
u
ti
o
n
was
p
r
es
er
v
e
d
ac
r
o
s
s
f
o
l
d
s
.
T
h
e
t
r
a
in
in
g
p
a
r
t
it
io
n
was
u
s
ed
to
tr
ai
n
b
ase
le
ar
n
e
r
s
i
n
e
ac
h
f
o
ld
,
wh
i
le
t
h
e
v
ali
d
ati
o
n
p
a
r
tit
io
n
was
u
s
e
d
t
o
r
e
co
r
d
t
h
e
ir
p
r
o
b
a
b
il
it
y
o
u
t
p
u
ts
.
A
n
e
w
f
ea
tu
r
e
s
p
a
ce
k
n
o
wn
as
t
h
e
m
et
a
-
f
e
at
u
r
es
was
t
h
e
n
p
r
o
d
u
ce
d
b
y
c
o
n
ca
t
e
n
at
in
g
t
h
es
e
O
OF
p
r
e
d
ic
ti
o
n
s
.
I
n
o
r
d
e
r
to
c
ap
t
u
r
e
t
h
e
r
el
ati
v
e
co
n
t
r
i
b
u
ti
o
n
s
o
f
th
e
b
as
e
c
las
s
if
i
er
s
a
n
d
est
ab
lis
h
t
h
e
b
est
wei
g
h
ts
f
o
r
t
h
ei
r
e
n
s
e
m
b
le
,
a
lo
g
is
t
ic
r
eg
r
ess
io
n
m
o
d
el
was
em
p
l
o
y
e
d
as
t
h
e
m
eta
-
le
a
r
n
er
.
T
h
e
a
g
g
r
eg
at
ed
p
r
ed
i
cti
o
n
s
f
r
o
m
al
l
f
o
l
d
s
w
er
e
t
h
e
n
u
s
e
d
t
o
ass
ess
th
e
f
in
al
s
tac
k
ed
m
o
d
el
o
n
th
e
ex
cl
u
d
e
d
tes
t
s
e
t.
B
y
u
s
i
n
g
t
h
e
OOF
s
t
r
a
te
g
y
,
th
is
m
et
h
o
d
r
e
d
u
ce
s
t
h
e
ch
a
n
ce
o
f
o
v
e
r
f
itti
n
g
wh
ile
u
til
izi
n
g
th
e
c
o
m
p
l
em
e
n
ta
r
y
ca
p
a
b
i
liti
es
o
f
a
h
ete
r
o
g
e
n
e
o
u
s
b
as
e.
2
.
7
.
P
er
f
o
r
m
a
nce
m
e
t
rics
T
h
e
h
y
b
r
i
d
m
o
d
el'
s
p
er
f
o
r
m
a
n
ce
was
ev
al
u
ated
u
s
in
g
a
n
u
m
b
er
o
f
ass
ess
m
en
t
cr
iter
ia.
Stan
d
ar
d
m
etr
ics
s
u
ch
as
p
r
ec
is
io
n
,
r
ec
all,
s
p
ec
if
icity
,
F1
-
s
co
r
e,
AUC,
an
d
ac
cu
r
ac
y
wer
e
u
s
ed
to
e
v
alu
ate
th
e
p
er
f
o
r
m
an
ce
o
f
m
ac
h
i
n
e
lear
n
in
g
class
if
ier
s
.
T
h
e
m
ath
em
ati
ca
l
r
ep
r
esen
tatio
n
s
o
f
t
h
ese
m
etr
ics
ar
e
g
iv
en
b
y
th
e
f
o
llo
win
g
f
o
r
m
u
las:
Acc
u
r
ac
y
=
(
TP
+
TN
)
(
+
+
+
)
(
4
)
R
ec
all
=
+
(
5
)
Pre
cisi
o
n
=
TP
(
TP
+
FP
)
(
6
)
Sp
ec
if
icity
=
TN
(
TN
+
FP
)
(
7
)
1
−
=
2
×
r
ec
al
l
×
p
r
ecis
i
o
n
+
(
8
)
T
h
e
AUC
s
co
r
e
r
an
g
es
f
r
o
m
0
to
1
,
an
d
th
e
m
o
d
el’
s
p
r
ed
ictiv
e
p
er
f
o
r
m
an
ce
im
p
r
o
v
es
as
th
e
v
alu
e
ap
p
r
o
ac
h
es 1
,
wh
er
ea
s
it d
ec
r
e
ases
as it g
ets clo
s
er
to
0
[
2
4
]
.
3.
RE
SU
L
T
S AN
D
D
I
SCU
SS
I
O
N
T
h
is
s
ec
tio
n
p
r
esen
ts
th
e
m
ain
f
in
d
in
g
s
o
f
th
e
s
tu
d
y
.
T
ab
le
2
s
u
m
m
ar
izes
th
e
cla
s
s
if
icatio
n
p
er
f
o
r
m
an
ce
o
f
t
h
e
f
o
u
r
m
o
d
els
o
n
th
e
h
eld
-
o
u
t
test
s
et.
A
m
o
n
g
th
e
i
n
d
iv
id
u
al
lear
n
e
r
s
,
XGBo
o
s
t
ac
h
iev
ed
th
e
b
est o
v
er
all
p
er
f
o
r
m
an
ce
,
with
an
ac
cu
r
ac
y
o
f
9
9
.
9
5
%,
p
r
ec
is
io
n
o
f
0
.
9
9
9
3
,
r
ec
all
o
f
0
.
9
9
9
9
,
s
p
ec
if
icity
o
f
0
.
9
9
9
1
,
a
n
d
F1
-
s
co
r
e
o
f
0
.
9
9
9
6
.
T
h
e
p
r
o
p
o
s
ed
s
tack
in
g
e
n
s
em
b
le,
wh
ich
c
o
m
b
in
es
th
e
OOF
p
r
ed
ictio
n
s
o
f
RF
,
XG
B
o
o
s
t
,
an
d
ML
P
th
r
o
u
g
h
a
lo
g
is
tic
-
r
eg
r
ess
io
n
m
et
a
-
lear
n
er
,
ac
h
iev
ed
th
e
h
ig
h
e
s
t
v
alu
es
ac
r
o
s
s
all
r
ep
o
r
ted
m
etr
ics,
in
clu
d
i
n
g
a
n
ac
cu
r
ac
y
o
f
9
9
.
9
7
%,
p
r
ec
i
s
io
n
o
f
0
.
9
9
9
4
,
r
ec
all
o
f
1
.
0
0
0
0
,
s
p
ec
if
icity
o
f
0
.
9
9
9
4
,
an
d
F1
-
s
co
r
e
o
f
0
.
9
9
9
7
.
C
o
m
p
ar
ed
with
XGBo
o
s
t
,
th
e
en
s
em
b
le
f
u
r
th
e
r
r
ef
in
ed
p
r
ed
ictiv
e
p
er
f
o
r
m
an
ce
,
im
p
r
o
v
in
g
ac
c
u
r
ac
y
b
y
0
.
0
2
p
e
r
ce
n
tag
e
p
o
i
n
ts
an
d
F1
-
s
co
r
e
b
y
0
.
0
0
0
1
,
wh
ile
ac
h
iev
in
g
a
p
er
f
ec
t
r
ec
all
o
f
1
0
0
.
0
0
%.
T
h
e
clo
s
e
alig
n
m
en
t
b
etwe
en
p
r
ec
is
io
n
an
d
r
ec
all
in
d
icate
s
a
h
ig
h
ly
b
alan
ce
d
class
if
icatio
n
p
er
f
o
r
m
a
n
ce
,
w
h
ile
th
e
co
n
s
is
ten
t
s
u
p
er
io
r
ity
o
v
er
all
in
d
iv
id
u
al
b
ase
lear
n
e
r
s
co
n
f
ir
m
s
th
at
th
e
s
tack
in
g
m
o
d
el
s
u
cc
ess
f
u
lly
lev
er
ag
es
co
m
p
lem
e
n
tar
y
p
r
ed
ictiv
e
in
f
o
r
m
atio
n
p
r
o
v
i
d
ed
b
y
t
h
e
b
ase
class
if
ier
s
.
W
ith
in
th
e
s
tack
in
g
f
r
am
ewo
r
k
,
a
5
-
f
o
ld
s
tr
atif
ied
cr
o
s
s
-
v
alid
atio
n
m
eth
o
d
was
u
s
ed
t
o
p
r
o
v
id
e
r
eliab
le
m
o
d
el
ass
ess
m
en
t
an
d
av
o
id
o
v
er
f
itti
n
g
.
E
ac
h
b
ase
l
ea
r
n
er
(
RF
,
XGBo
o
s
t,
an
d
M
L
P)
p
r
o
d
u
ce
d
OOF
p
r
ed
ictio
n
s
,
wh
ich
wer
e
th
en
f
ed
in
to
th
e
m
eta
-
lear
n
e
r
.
T
h
is
m
eth
o
d
en
h
a
n
ce
s
g
en
er
ali
za
tio
n
p
er
f
o
r
m
an
ce
an
d
g
u
ar
an
tees th
at
th
e
m
eta
-
m
o
d
el
is
tr
ain
ed
o
n
o
b
jectiv
e
p
r
ed
ictio
n
s
.
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
A
n
a
r
tifi
cia
l in
tellig
en
ce
-
b
a
s
ed
s
ta
ck
in
g
en
s
emb
le
fr
a
mewo
r
k
fo
r
s
ma
r
t i
r
r
ig
a
tio
n
…
(
S
a
r
r
a
Go
u
r
a
r
i
)
2657
T
ab
le
2
.
Per
f
o
r
m
an
ce
r
esu
lts
o
f
class
if
icatio
n
p
er
f
o
r
m
a
n
ce
s
M
o
d
e
l
s
P
r
e
c
i
s
i
o
n
F1
-
sc
o
r
e
R
e
c
a
l
l
S
p
e
c
i
f
i
c
i
t
y
A
c
c
u
r
a
c
y
(
%)
RF
0
.
9
9
7
9
0
.
9
9
8
7
0
.
9
9
9
6
0
.
9
9
7
5
9
9
.
8
6
M
LP
0
.
9
6
7
6
0
.
9
7
5
5
0
.
9
8
3
5
0
.
9
6
1
8
9
7
.
3
4
X
G
B
o
o
st
0
.
9
9
9
3
0
.
9
9
9
6
0
.
9
9
9
9
0
.
9
9
9
1
9
9
.
9
5
S
t
a
c
k
i
n
g
0
.
9
9
9
4
0
.
9
9
9
7
1
.
0
0
0
0
0
.
9
9
9
4
9
9
.
9
7
3
.
1
.
St
a
t
is
t
ica
l
v
a
lid
a
t
io
n
a
nd
o
v
er
f
it
t
ing
a
s
s
ess
m
ent
T
o
q
u
an
tify
esti
m
ate
s
tab
ilit
y
,
5
-
f
o
ld
s
tr
atif
ied
cr
o
s
s
-
v
alid
a
tio
n
was
ad
d
itio
n
ally
p
er
f
o
r
m
ed
o
n
th
e
tr
ain
in
g
s
p
lit
f
o
r
e
ac
h
b
ase
le
ar
n
er
u
s
in
g
its
s
elec
ted
h
y
p
er
p
ar
am
eter
s
(
T
ab
le
1
)
,
a
n
d
th
e
f
o
ld
-
to
-
f
o
ld
m
ea
n
an
d
s
tan
d
ar
d
d
ev
iatio
n
ar
e
r
e
p
o
r
ted
f
o
r
e
v
er
y
m
etr
ic
in
T
a
b
le
3
.
Stan
d
ar
d
d
e
v
iatio
n
s
r
e
m
ain
ed
b
elo
w
0
.
0
0
3
f
o
r
all
m
etr
ics
an
d
all
m
o
d
el
s
,
in
d
icatin
g
s
tab
le,
lo
w
-
v
a
r
ian
ce
esti
m
ates
r
ath
er
t
h
an
a
r
esu
lt
d
r
iv
en
b
y
a
f
av
o
u
r
ab
le
s
in
g
le
s
p
lit.
Data
leak
ag
e
was p
r
ev
en
ted
b
y
p
er
f
o
r
m
in
g
th
e
tr
ai
n
/tes
t sp
li
t p
r
io
r
to
an
y
p
r
ep
r
o
ce
s
s
in
g
an
d
b
y
f
itti
n
g
th
e
f
ea
tu
r
e
s
ca
ler
ex
clu
s
iv
ely
o
n
th
e
tr
ain
in
g
p
ar
titi
o
n
.
T
ab
l
e
4
r
ep
o
r
ts
th
e
tr
ain
-
test
ac
cu
r
ac
y
g
ap
f
o
r
ev
e
r
y
m
o
d
el;
all
g
a
p
s
r
em
ain
b
elo
w
0
.
5
p
er
ce
n
ta
g
e
p
o
in
ts
,
wh
ich
we
r
ep
o
r
t
as
d
ir
ec
t
q
u
an
titativ
e
ev
id
en
ce
ag
ain
s
t
class
ical
o
v
er
f
itti
n
g
.
T
h
e
v
e
r
y
h
ig
h
ab
s
o
lu
te
ac
c
u
r
ac
y
is
d
is
cu
s
s
ed
s
ep
ar
ately
in
s
ec
tio
n
3
.
3
.
T
ab
le
3
.
5
-
f
o
ld
cr
o
s
s
-
v
alid
atio
n
s
tatis
tical
v
alid
atio
n
o
n
th
e
t
r
ain
in
g
s
p
lit (
m
ea
n
±
s
td
)
M
e
t
r
i
c
RF
X
G
B
o
o
st
M
LP
A
c
c
u
r
a
c
y
0
.
9
9
8
1
±
0
.
0
0
0
3
0
.
9
9
9
5
±
0
.
0
0
0
1
0
.
9
6
9
7
±
0
.
0
0
2
1
P
r
e
c
i
s
i
o
n
0
.
9
9
7
3
±
0
.
0
0
0
4
0
.
9
9
9
4
±
0
.
0
0
0
1
0
.
9
6
3
3
±
0
.
0
0
2
0
R
e
c
a
l
l
0
.
9
9
9
3
±
0
.
0
0
0
4
0
.
9
9
9
6
±
0
.
0
0
0
2
0
.
9
8
0
9
±
0
.
0
0
2
5
F1
-
sc
o
r
e
0
.
9
9
8
3
±
0
.
0
0
0
2
0
.
9
9
9
5
±
0
.
0
0
0
1
0
.
9
7
2
0
±
0
.
0
0
1
9
S
p
e
c
i
f
i
c
i
t
y
0
.
9
9
6
8
±
0
.
0
0
0
5
0
.
9
9
9
4
±
0
.
0
0
0
1
0
.
9
5
6
7
±
0
.
0
0
2
4
AUC
1
.
0
0
0
0
±
0
.
0
0
0
0
1
.
0
0
0
0
±
0
.
0
0
0
0
0
.
9
9
6
1
±
0
.
0
0
0
7
T
ab
le
4
.
T
r
ai
n
-
test
ac
cu
r
ac
y
g
ap
p
er
m
o
d
el
(
o
v
er
f
itti
n
g
ch
ec
k
)
M
o
d
e
l
Tr
a
i
n
a
c
c
u
r
a
c
y
(
%)
Te
st
a
c
c
u
r
a
c
y
(
%)
Tr
a
i
n
-
t
e
s
t
g
a
p
(
p
p
)
RF
1
0
0
.
0
0
9
9
.
8
6
0
.
1
4
X
G
B
o
o
st
9
9
.
9
9
9
9
.
9
5
0
.
0
4
M
LP
9
7
.
7
6
9
7
.
3
4
0
.
4
2
S
t
a
c
k
i
n
g
1
0
0
.
0
0
9
9
.
9
7
0
.
0
3
3
.
2
.
Why
t
he
s
t
a
c
k
ing
ens
em
ble o
utper
f
o
rm
s
t
he
ba
s
e
l
ea
rner
s
B
ey
o
n
d
r
ep
o
r
tin
g
th
e
ag
g
r
eg
ate
ac
cu
r
ac
y
g
ain
,
th
e
p
air
w
is
e
p
r
ed
ictio
n
ag
r
ee
m
en
t
b
et
wee
n
b
ase
lear
n
er
s
o
n
t
h
e
h
eld
-
o
u
t
test
s
et
was
co
m
p
u
ted
to
q
u
a
n
tify
h
o
w
m
u
ch
c
o
m
p
lem
e
n
tar
y
in
f
o
r
m
atio
n
is
av
ailab
le
to
th
e
m
eta
-
lear
n
er
as
s
h
o
wn
in
T
ab
le
5
.
RF
an
d
XGBo
o
s
t
ag
r
ee
o
n
9
9
.
8
6
%
o
f
test
in
s
tan
ce
s
,
wh
ile
ML
P
ag
r
ee
s
with
ea
ch
tr
ee
-
b
ased
le
ar
n
er
o
n
ap
p
r
o
x
im
ately
9
7
.
4
%
o
f
in
s
tan
ce
s
.
Alth
o
u
g
h
th
ese
ag
r
ee
m
en
t
r
ates
ar
e
h
ig
h
in
ab
s
o
lu
te
ter
m
s
,
th
e
r
esid
u
al
d
is
ag
r
ee
m
en
t is n
o
t r
an
d
o
m
: b
ec
au
s
e
th
e
th
r
ee
b
ase
lear
n
er
s
h
av
e
d
if
f
e
r
en
t
in
d
u
ctiv
e
b
iases
,
th
ey
ten
d
to
m
ak
e
er
r
o
r
s
o
n
d
if
f
er
e
n
t
s
u
b
s
ets
o
f
in
s
tan
ce
s
.
I
t
is
p
r
ec
is
e
ly
th
is
s
m
all,
n
o
n
-
o
v
er
lap
p
i
n
g
r
esid
u
al
d
is
ag
r
ee
m
en
t
th
at
th
e
lo
g
is
tic
-
r
eg
r
ess
io
n
m
eta
-
lear
n
er
e
x
p
lo
its
,
wh
i
ch
ex
p
lain
s
wh
y
th
e
Stack
in
g
en
s
em
b
le
im
p
r
o
v
es
o
v
er
ev
en
th
e
s
tr
o
n
g
est
in
d
i
v
id
u
al
b
ase
lear
n
er
(
XGBo
o
s
t
)
d
esp
ite
th
e
h
ig
h
p
air
wis
e
ag
r
ee
m
en
t b
etwe
en
t
h
e
b
ase
lear
n
er
s
.
T
ab
le
5
.
Pair
wis
e
b
ase
-
lear
n
er
p
r
ed
ictio
n
a
g
r
ee
m
e
n
t o
n
t
h
e
h
eld
-
o
u
t te
s
t set
M
o
d
e
l
A
M
o
d
e
l
B
P
r
e
d
i
c
t
i
o
n
a
g
r
e
e
m
e
n
t
(
%)
RF
X
G
B
o
o
st
9
9
.
8
6
RF
M
LP
9
7
.
4
0
X
G
B
o
o
st
M
LP
9
7
.
3
6
Fig
u
r
e
2
p
r
esen
ts
th
e
co
n
f
u
s
io
n
m
atr
ices
f
o
r
th
e
ev
alu
ate
d
m
o
d
els
o
n
th
e
test
s
et
(
n
=
1
9
,
9
6
0
)
.
W
h
ile
ML
P
ex
h
ib
ited
th
e
h
ig
h
est
n
u
m
b
er
o
f
m
is
class
if
icatio
n
s
(
5
3
0
to
tal
er
r
o
r
s
:
3
5
3
f
alse
p
o
s
itiv
es
an
d
1
7
7
f
alse
n
eg
ativ
es),
RF
an
d
XGBo
o
s
t
d
em
o
n
s
tr
ated
s
ig
n
if
ican
tly
s
tr
o
n
g
er
p
er
f
o
r
m
a
n
ce
,
r
ed
u
ci
n
g
to
tal
er
r
o
r
s
to
2
7
an
d
9
,
r
esp
ec
tiv
ely
.
T
h
e
p
r
o
p
o
s
ed
s
tack
in
g
en
s
em
b
le
ac
h
iev
e
d
th
e
b
est
o
v
er
all
r
esu
lts
,
p
r
o
d
u
cin
g
o
n
ly
6
f
alse
p
o
s
itiv
es
an
d
0
f
alse
n
eg
ati
v
es
—
y
ield
in
g
an
e
x
ce
p
tio
n
al
ly
lo
w
o
v
e
r
all
m
is
class
if
icati
o
n
r
ate
o
f
0
.
0
3
%.
No
tab
ly
,
th
e
s
tack
in
g
e
n
s
em
b
le
elim
in
ated
f
alse
n
eg
ativ
es
(
0
m
is
s
ed
ac
tiv
atio
n
s
)
,
d
em
o
n
s
tr
atin
g
an
ab
s
o
lu
te
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
6
5
2
-
2
6
6
3
2658
ab
ilit
y
to
a
v
o
id
m
is
s
ed
ir
r
ig
at
io
n
tr
ig
g
er
s
.
T
h
is
p
er
f
o
r
m
an
c
e
is
o
p
er
atio
n
ally
c
r
itical,
as
a
m
is
s
ed
ac
tiv
atio
n
(
f
alse n
eg
ativ
e)
r
is
k
s
cr
o
p
wat
er
s
tr
ess
an
d
s
ev
er
e
y
ield
lo
s
s
,
wh
er
ea
s
a
f
alse a
ctiv
atio
n
(
f
alse p
o
s
itiv
e)
m
er
ely
r
esu
lts
in
m
in
o
r
ex
ce
s
s
wate
r
u
s
ag
e.
Fig
u
r
e
3
p
r
es
en
ts
R
OC
c
u
r
v
es
o
f
R
F
,
X
GB
o
o
s
t
(
XGB
)
,
a
n
d
M
eta
-
L
ea
r
n
e
r
m
o
d
els
a
n
d
t
h
ei
r
class
i
f
i
ca
t
io
n
p
e
r
f
o
r
m
a
n
c
e.
T
h
e
R
F
m
o
d
el
s
h
o
we
d
g
o
o
d
p
r
ed
icti
v
e
p
o
we
r
w
it
h
a
n
AUC
o
f
0
.
9
7
.
T
h
e
XGB
o
o
s
t
m
o
d
el
p
er
f
o
r
m
e
d
s
o
m
ew
h
at
b
ette
r
,
s
h
o
wi
n
g
im
p
r
o
v
ed
s
e
n
s
i
t
iv
i
ty
a
n
d
ac
c
u
r
ac
y
wit
h
a
n
AU
C
o
f
0
.
9
8
.
W
it
h
a
n
AUC
o
f
0
.
9
9
,
th
e
Me
t
a
-
L
e
a
r
n
e
r
m
o
d
el
f
a
r
e
d
b
ett
er
t
h
a
n
b
o
th
,
d
em
o
n
s
t
r
a
ti
n
g
i
ts
s
u
p
e
r
i
o
r
ca
p
ac
it
y
t
o
g
e
n
e
r
al
ize
an
d
s
u
c
ce
s
s
f
u
ll
y
i
n
t
eg
r
a
te
th
e
ad
v
a
n
t
a
g
es
o
f
b
ase
m
o
d
els
.
T
h
es
e
f
in
d
i
n
g
s
d
e
m
o
n
s
tr
ate
t
h
a
t
th
e
m
et
a
-
le
ar
n
i
n
g
en
s
em
b
l
e
o
u
t
p
e
r
f
o
r
m
s
i
n
d
iv
i
d
u
al
m
o
d
els
b
y
p
r
ese
r
v
in
g
g
r
ea
te
r
t
r
u
e
p
o
s
i
ti
v
e
r
a
tes
w
h
il
e
r
e
d
u
ci
n
g
f
a
ls
e
p
o
s
iti
v
es
,
a
n
d
t
h
at
it
c
o
n
s
is
te
n
t
ly
p
r
o
d
u
ce
s
t
h
e
m
o
s
t
a
cc
u
r
at
e
an
d
d
ep
e
n
d
a
b
le
cl
ass
i
f
ic
ati
o
n
r
esu
l
ts
.
Fig
u
r
e
2
.
C
o
n
f
u
s
io
n
m
atr
ices
o
f
class
if
icatio
n
alg
o
r
ith
m
s
: RF
,
ML
P,
XGBo
o
s
t,
an
d
s
tack
in
g
Fig
u
r
e
3
.
R
OC
c
u
r
v
es
a
n
d
AU
C
v
al
u
es
c
o
m
p
ar
i
n
g
m
o
d
el
p
e
r
f
o
r
m
a
n
c
es
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
A
n
a
r
tifi
cia
l in
tellig
en
ce
-
b
a
s
ed
s
ta
ck
in
g
en
s
emb
le
fr
a
mewo
r
k
fo
r
s
ma
r
t i
r
r
ig
a
tio
n
…
(
S
a
r
r
a
Go
u
r
a
r
i
)
2659
T
h
e
R
F
a
n
d
XGB
o
o
s
t
m
o
d
el
s
b
o
t
h
ac
h
i
e
v
e
d
a
n
AUC
o
f
1
.
0
0
0
0
,
a
n
d
t
h
e
s
tac
k
i
n
g
m
eta
-
le
a
r
n
er
lik
ewis
e
ac
h
i
ev
e
d
a
n
A
UC
o
f
1
.
0
0
0
0
,
w
h
il
e
th
e
ML
P
m
o
d
el
a
ch
ie
v
e
d
a
c
o
m
p
a
r
ati
v
el
y
lo
w
er
AUC
o
f
0
.
9
9
7
2
.
T
h
e
n
ea
r
-
p
er
f
e
ct
AUC
v
al
u
es
f
o
r
t
h
e
t
r
e
e
-
b
ase
d
a
n
d
s
tac
k
i
n
g
m
o
d
els
m
ir
r
o
r
th
e
v
er
y
l
o
w
m
is
c
lass
i
f
i
ca
t
io
n
co
u
n
ts
a
lr
ea
d
y
r
e
p
o
r
t
ed
f
o
r
t
h
ese
m
o
d
e
ls
i
n
Fi
g
u
r
e
2
,
an
d
a
r
e
ex
am
in
e
d
f
u
r
t
h
e
r
,
to
g
et
h
e
r
wit
h
t
h
e
ass
o
cia
te
d
o
v
e
r
f
itti
n
g
c
h
ec
k
s
a
n
d
t
h
e
r
e
c
o
m
m
en
d
ed
e
x
t
er
n
al
-
d
atas
et
v
ali
d
at
io
n
,
in
s
e
cti
o
n
3
.
1
.
Ou
r
s
t
ac
k
i
n
g
e
n
s
e
m
b
le
m
o
d
el
is
c
o
m
p
a
r
e
d
wi
th
a
n
u
m
b
e
r
o
f
c
u
r
r
e
n
t
s
tat
e
-
of
-
t
h
e
-
a
r
t
te
c
h
n
i
q
u
es,
as
in
d
ic
at
ed
i
n
T
ab
le
6
.
Usi
n
g
a
d
at
ase
t
t
h
at
c
o
m
b
i
n
es
ac
t
u
al
I
o
T
s
e
n
s
o
r
r
e
ad
in
g
s
wi
th
p
u
b
li
cly
av
ail
ab
le
Ka
g
g
l
e
d
at
a,
i
n
cl
u
d
i
n
g
c
h
ar
ac
t
er
is
ti
cs
li
k
e
te
m
p
er
at
u
r
e,
h
u
m
i
d
it
y
,
a
n
d
s
o
il
m
o
is
t
u
r
e
,
o
u
r
h
y
b
r
i
d
m
o
d
el
s
h
o
ws
h
i
g
h
e
r
p
e
r
f
o
r
m
a
n
c
e.
C
o
m
p
a
r
ed
to
m
o
d
els
t
r
ai
n
e
d
o
n
cu
r
a
te
d
o
r
les
s
d
i
v
e
r
s
e
d
a
tas
ets,
th
e
le
a
r
n
in
g
c
h
a
lle
n
g
e
is
m
o
r
e
d
i
f
f
ic
u
lt
d
u
e
to
t
h
e
d
i
v
e
r
s
it
y
a
n
d
r
e
al
-
w
o
r
ld
c
h
ar
ac
t
er
o
f
t
h
e
d
atas
et.
T
a
b
le
6
.
A
cc
u
r
ac
y
c
o
m
p
a
r
is
o
n
o
f
o
u
r
p
r
o
p
o
s
e
d
s
t
u
d
y
wi
th
s
i
m
il
ar
li
te
r
at
u
r
e
s
t
u
d
ies
Ref
Y
e
a
r
s
Te
c
h
n
i
q
u
e
F
e
a
t
u
r
e
s
A
c
c
u
r
a
c
y
[
2
5
]
2
0
2
2
D
T,
R
F
,
a
n
d
S
V
M
S
o
i
l
m
o
i
s
t
u
r
e
,
t
e
m
p
e
r
a
t
u
r
e
,
h
u
mi
d
i
t
y
,
t
i
m
e
9
5
%
[
2
6
]
2
0
2
4
ANN
S
o
i
l
m
o
i
s
t
u
r
e
,
t
e
m
p
e
r
a
t
u
r
e
,
h
u
mi
d
i
t
y
9
7
.
6
%
[
2
7
]
2
0
2
1
D
e
c
i
s
i
o
n
Tr
e
e
S
o
i
l
m
o
i
s
t
u
r
e
,
t
e
m
p
e
r
a
t
u
r
e
9
7
.
8
6
%
[
1
8
]
2
0
2
2
C
N
N
S
o
i
l
m
o
i
s
t
u
r
e
,
t
e
m
p
e
r
a
t
u
r
e
,
h
u
mi
d
i
t
y
,
t
i
m
e
9
8
.
2
3
%
[
2
8
]
2
0
2
4
K
-
me
a
n
s
+
S
V
M
(
H
y
b
r
i
d
)
S
o
i
l
m
o
i
s
t
u
r
e
,
t
e
m
p
e
r
a
t
u
r
e
,
h
u
mi
d
i
t
y
9
8
.
5
%
P
r
o
p
o
se
d
m
o
d
e
l
S
t
a
c
k
i
n
g
(
RF
,
X
G
B
o
o
st
,
M
LP)
S
o
i
l
m
o
i
s
t
u
r
e
,
t
e
m
p
e
r
a
t
u
r
e
,
h
u
mi
d
i
t
y
,
t
i
m
e
9
9
.
9
7
%
3
.
3
.
M
o
del
i
nte
rpre
t
a
bil
it
y
:
SH
AP
a
na
ly
s
is
T
o
m
o
v
e
b
ey
o
n
d
a
p
u
r
e
ly
p
r
ed
i
cti
v
e
e
v
a
l
u
ati
o
n
,
SHAP
w
as
u
s
e
d
t
o
in
te
r
p
r
et
t
h
e
m
o
d
e
l'
s
d
e
cisi
o
n
d
r
iv
e
r
s
.
B
e
ca
u
s
e
ef
f
i
cie
n
t
tr
ee
-
b
ase
d
SHAP
e
x
p
la
n
at
io
n
(
T
r
e
eE
x
p
la
in
er
)
is
n
o
t
n
a
ti
v
el
y
d
e
f
i
n
e
d
f
o
r
a
h
et
e
r
o
g
e
n
eo
u
s
s
ta
ck
in
g
cl
ass
if
ier
,
th
e
r
e
p
o
r
te
d
SH
AP
v
a
lu
es
e
x
p
lai
n
th
e
X
GB
o
o
s
t
b
ase
le
a
r
n
e
r
,
wh
ic
h
is
b
o
th
th
e
b
est
-
p
e
r
f
o
r
m
i
n
g
i
n
d
i
v
i
d
u
a
l
m
o
d
e
l
a
n
d
,
g
i
v
e
n
its
9
7
.
4
%
–
9
9
.
9
%
p
r
e
d
ic
ti
o
n
a
g
r
e
em
en
t
wit
h
t
h
e
o
t
h
e
r
b
as
e
lea
r
n
e
r
s
as
s
h
o
w
n
i
n
T
a
b
l
e
5
,
r
e
p
r
ese
n
t
ati
v
e
o
f
t
h
e
d
ec
is
i
o
n
d
r
iv
er
s
u
s
e
d
a
cr
o
s
s
t
h
e
e
n
s
em
b
le
as
a
w
h
o
le
.
Fig
u
r
e
4
s
h
o
ws
t
h
e
S
HAP
s
u
m
m
ar
y
(
b
e
eswa
r
m
)
p
l
o
t:
T
e
m
p
e
r
at
u
r
e
a
n
d
t
h
e
c
y
c
li
ca
ll
y
-
e
n
c
o
d
e
d
ti
m
e
co
m
p
o
n
e
n
ts
(
T
i
m
e
_
s
i
n
,
T
i
m
e
_
c
o
s
)
a
r
e
t
h
e
d
o
m
in
an
t
d
r
iv
er
s
o
f
t
h
e
p
r
e
d
i
cte
d
p
u
m
p
s
tat
u
s
,
f
o
l
lo
we
d
b
y
s
o
il
m
o
is
t
u
r
e
a
n
d
h
u
m
i
d
i
ty
;
co
lo
r
e
n
c
o
d
es
e
ac
h
i
n
s
ta
n
c
e'
s
o
r
ig
in
a
l
f
e
at
u
r
e
v
al
u
e
,
s
o
t
h
e
p
l
o
t
c
o
n
v
e
y
s
b
o
t
h
t
h
e
m
a
g
n
it
u
d
e
a
n
d
t
h
e
d
i
r
e
cti
o
n
o
f
e
v
e
r
y
f
e
at
u
r
e'
s
e
f
f
ec
t,
n
o
t
o
n
l
y
its
r
a
n
k
in
g
.
Fig
u
r
e
4
.
SHAP su
m
m
ar
y
(
b
e
eswar
m
)
p
lo
t f
o
r
th
e
XGBo
o
s
t b
ase
lear
n
er
o
n
th
e
h
eld
-
o
u
t t
est s
et
Fig
u
r
e
5
p
r
esen
ts
SHAP
d
ep
e
n
d
en
ce
p
lo
ts
f
o
r
t
h
e
two
m
o
s
t
in
f
lu
en
tial
f
ea
tu
r
es.
Pan
el
(
a)
s
h
o
ws
th
at
th
e
SHAP
v
alu
e
f
o
r
T
em
p
e
r
a
tu
r
e
in
cr
ea
s
es
m
o
n
o
to
n
ically
with
tem
p
er
atu
r
e,
m
ea
n
in
g
h
i
g
h
er
tem
p
er
atu
r
es
co
n
s
is
ten
tly
p
u
s
h
th
e
m
o
d
el
to
war
d
p
r
ed
ictin
g
p
u
m
p
ac
tiv
a
tio
n
(
ON)
—
a
p
h
y
s
ically
p
la
u
s
ib
le
r
elatio
n
s
h
ip
,
co
n
s
is
ten
t
with
h
ig
h
er
ev
ap
o
t
r
an
s
p
ir
atio
n
an
d
ir
r
ig
atio
n
d
e
m
an
d
u
n
d
er
war
m
er
co
n
d
itio
n
s
.
Pan
el
(
b
)
s
h
o
ws
two
d
is
tin
ct
clu
s
ter
s
f
o
r
th
e
T
i
m
e_
co
s
f
ea
tu
r
e,
co
r
r
esp
o
n
d
in
g
to
d
if
f
er
e
n
t
p
h
ases
o
f
th
e
d
a
ily
cy
cle,
in
d
icatin
g
th
at
p
r
ed
icted
p
u
m
p
-
ac
tiv
atio
n
lik
elih
o
o
d
v
ar
ies
s
y
s
tem
atica
lly
with
tim
e
-
of
-
d
ay
b
ey
o
n
d
wh
at
tem
p
er
atu
r
e
alo
n
e
ex
p
lain
s
—
s
u
p
p
o
r
tin
g
th
e
ch
o
ice
o
f
c
y
clica
l
(
s
in
e
-
co
s
in
e)
en
co
d
i
n
g
o
v
e
r
a
r
aw
lin
ea
r
tim
e
f
ea
tu
r
e.
B
o
th
tr
en
d
s
b
ein
g
p
h
y
s
ically
in
ter
p
r
etab
le,
r
ath
er
th
an
er
r
atic
o
r
r
e
v
er
s
ed
,
p
r
o
v
id
e
a
d
d
itio
n
al
q
u
alitativ
e
ev
id
en
ce
th
at
th
e
m
o
d
els
a
r
e
lear
n
in
g
a
g
en
u
in
e,
p
h
y
s
ically
g
r
o
u
n
d
ed
s
ig
n
al
r
ath
e
r
th
an
a
s
p
u
r
io
u
s
co
r
r
elatio
n
,
wh
ic
h
is
r
elev
an
t
to
th
e
v
er
y
h
ig
h
ac
cu
r
ac
y
d
i
s
cu
s
s
ed
in
s
ec
tio
n
3
.
1
.
T
ab
le
7
r
ep
o
r
ts
th
e
ex
ac
t
r
an
k
ed
m
ea
n
ab
s
o
lu
te
SHAP v
alu
es
u
n
d
er
ly
in
g
th
e
v
is
u
al
o
r
d
er
in
g
s
h
o
wn
in
Fig
u
r
e
4
,
p
r
o
v
id
in
g
r
ep
r
o
d
u
ci
b
le
n
u
m
er
ic
v
al
u
es f
o
r
c
o
m
p
ar
is
o
n
in
f
u
t
u
r
e
wo
r
k
.
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
6
5
2
-
2
6
6
3
2660
(
a)
(
b
)
Fig
u
r
e
5
.
SHAP d
ep
en
d
en
ce
p
lo
ts
f
o
r
(
a)
T
em
p
er
atu
r
e
an
d
(
b
)
T
im
e_
c
o
s
,
co
lo
r
e
d
b
y
th
e
s
ec
o
n
d
ar
y
f
ea
tu
r
e
s
h
o
wn
i
n
ea
ch
c
o
lo
r
b
a
r
T
ab
le
7
.
R
an
k
e
d
g
lo
b
al
f
ea
tu
r
e
im
p
o
r
tan
ce
(
m
ea
n
a
b
s
o
lu
te
SHAP v
alu
e)
,
XGBo
o
s
t b
ase
l
ea
r
n
er
,
h
eld
-
o
u
t te
s
t set
F
e
a
t
u
r
e
M
e
a
n
|
S
H
A
P
v
a
l
u
e
|
Te
mp
e
r
a
t
u
r
e
3
.
0
7
5
Ti
me
c
o
s
2
.
5
0
9
S
o
i
l
M
o
i
st
u
r
e
2
.
3
0
1
Ti
me
si
n
1
.
7
8
4
H
u
mi
d
i
t
y
1
.
5
8
9
3
.
4
.
Co
m
pu
t
a
t
io
na
l
c
o
m
plex
it
y
a
nd
edg
e
-
deplo
y
m
ent
f
e
a
s
ibi
lity
Giv
en
th
e
I
o
T
c
o
n
tex
t
o
f
th
is
ap
p
licatio
n
,
p
e
r
-
m
o
d
el
tr
ain
in
g
tim
e,
in
f
er
en
c
e
laten
cy
,
th
r
o
u
g
h
p
u
t,
an
d
m
em
o
r
y
f
o
o
tp
r
in
t
wer
e
m
ea
s
u
r
ed
o
n
th
e
h
eld
-
o
u
t
test
s
et
(
T
ab
le
8
)
.
XGBo
o
s
t
o
f
f
er
s
th
e
m
o
s
t
f
av
o
u
r
a
b
le
ac
cu
r
ac
y
-
to
-
co
s
t
r
atio
(
0
.
0
0
2
m
s
/s
am
p
le,
6
4
1
KB
)
,
m
a
k
in
g
it
in
d
iv
i
d
u
ally
well
-
s
u
ited
to
co
n
s
tr
ain
ed
m
icr
o
co
n
tr
o
ller
-
class
h
ar
d
war
e.
T
h
e
Stack
in
g
en
s
em
b
le'
s
in
f
er
en
ce
co
s
t
is
ap
p
r
o
x
im
ately
th
e
s
u
m
o
f
its
b
ase
lear
n
er
s
'
co
s
ts
p
lu
s
a
s
m
all
lo
g
is
tic
-
r
eg
r
ess
io
n
o
v
er
h
ea
d
(
0
.
0
0
9
m
s
/s
am
p
le)
,
an
d
its
m
em
o
r
y
f
o
o
tp
r
in
t
(
4
3
.
6
MB)
is
s
u
b
s
tan
tially
lar
g
er
th
a
n
an
y
in
d
iv
id
u
al
b
ase
lea
r
n
er
.
W
e
th
er
ef
o
r
e
r
ec
o
m
m
en
d
th
e
Stack
in
g
en
s
em
b
le
f
o
r
g
atew
ay
-
class
ed
g
e
h
ar
d
war
e
wh
er
e
its
m
ar
g
in
al
ac
cu
r
ac
y
an
d
r
ec
all
ad
v
an
tag
e
s
ar
e
o
p
er
atio
n
ally
v
alu
ab
le
(
s
ec
tio
n
3
.
1
)
,
a
n
d
r
ec
o
m
m
en
d
th
e
s
tan
d
al
o
n
e
XGBo
o
s
t
m
o
d
el
wh
er
e
h
ar
d
war
e
is
h
ig
h
ly
co
n
s
tr
ain
e
d
,
as
th
e
ac
cu
r
ac
y
d
if
f
er
en
ce
b
e
twee
n
th
e
two
(
0
.
0
2
p
er
ce
n
ta
g
e
p
o
in
ts
)
is
s
m
all
r
elativ
e
t
o
th
e
d
if
f
er
en
ce
in
d
ep
lo
y
m
e
n
t c
o
s
t.
T
ab
le
8
.
C
o
m
p
u
tatio
n
al
co
m
p
l
ex
ity
an
d
e
d
g
e
-
d
ep
lo
y
m
en
t f
e
asib
ilit
y
p
er
m
o
d
el
(
h
eld
-
o
u
t t
est s
et)
M
o
d
e
l
I
n
f
e
r
e
n
c
e
t
i
me
(
ms/sa
mp
l
e
)
Th
r
o
u
g
h
p
u
t
(
sam
p
l
e
s/
s)
M
o
d
e
l
s
i
z
e
(
K
B
)
M
LP
0
.
0
0
1
5
6
6
3
,
3
8
4
7
0
.
5
X
G
B
o
o
st
0
.
0
0
2
0
4
9
2
,
3
8
4
6
4
1
.
2
RF
0
.
0
0
6
3
1
5
9
,
5
2
3
2
1
,
0
8
9
.
8
S
t
a
c
k
i
n
g
0
.
0
0
9
3
1
0
7
,
5
2
3
4
3
,
6
0
8
.
6
3
.
5
.
E
x
t
er
na
l
da
t
a
s
et
v
a
lid
a
t
io
n
T
o
d
ir
ec
tly
ad
d
r
ess
co
n
ce
r
n
s
r
eg
ar
d
in
g
th
e
v
e
r
y
h
ig
h
in
-
d
o
m
ain
ac
cu
r
ac
y
,
th
e
i
d
en
tical
m
o
d
ellin
g
p
ip
elin
e
—
in
clu
d
in
g
m
o
d
el
ar
ch
itectu
r
e,
th
e
5
-
f
o
ld
h
y
p
er
p
ar
am
eter
-
s
ea
r
c
h
p
r
o
to
co
l,
an
d
th
e
ev
alu
atio
n
p
r
o
ce
d
u
r
e
—
was
r
e
-
ap
p
lied
,
with
o
u
t
r
eu
s
e
o
f
th
e
o
r
ig
in
al
f
itted
weig
h
ts
,
to
an
in
d
ep
en
d
e
n
t
d
ataset
co
llected
at
a
d
if
f
er
e
n
t
in
s
titu
tio
n
u
s
in
g
d
if
f
er
e
n
t
s
en
s
o
r
s
[
2
9
]
,
c
o
m
p
r
is
in
g
3
,
0
0
0
s
am
p
les
an
d
la
ck
in
g
a
c
o
m
p
ar
a
b
le
tem
p
o
r
al
f
ea
tu
r
e.
On
a
h
el
d
-
o
u
t
2
0
%
s
p
lit
o
f
th
is
ex
ter
n
al
d
ataset
(
6
0
0
s
am
p
les),
RF
,
XG
B
o
o
s
t
,
an
d
Stack
in
g
ac
h
iev
ed
1
0
0
.
0
0
%
ac
cu
r
ac
y
,
a
n
d
ML
P
ac
h
iev
ed
9
9
.
8
3
%
ac
c
u
r
ac
y
(
T
a
b
le
9
)
.
T
h
is
r
esu
lt
s
h
o
u
ld
b
e
i
n
ter
p
r
eted
with
ap
p
r
o
p
r
iate
ca
u
tio
n
:
g
iv
en
th
e
r
elativ
ely
s
m
all
s
iz
e
o
f
th
e
ex
te
r
n
al
test
s
et
(
n
=
6
0
0
)
,
a
s
in
g
l
e
m
is
class
if
icatio
n
wo
u
ld
alr
ea
d
y
r
ed
u
ce
ac
cu
r
ac
y
b
elo
w
1
0
0
%,
an
d
th
e
n
ea
r
-
p
e
r
f
ec
t sep
ar
a
b
ilit
y
o
b
s
er
v
ed
h
e
r
e
—
co
n
s
is
ten
t
with
th
e
p
atter
n
alr
ea
d
y
d
is
cu
s
s
ed
f
o
r
th
e
p
r
i
m
ar
y
d
ataset
—
is
co
m
p
atib
l
e
with
a
th
r
esh
o
ld
-
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
A
n
a
r
tifi
cia
l in
tellig
en
ce
-
b
a
s
ed
s
ta
ck
in
g
en
s
emb
le
fr
a
mewo
r
k
fo
r
s
ma
r
t i
r
r
ig
a
tio
n
…
(
S
a
r
r
a
Go
u
r
a
r
i
)
2661
b
ased
,
clo
s
e
-
to
-
d
eter
m
in
is
tic
s
en
s
o
r
-
to
-
lab
el
r
elatio
n
s
h
ip
ch
ar
ac
ter
is
tic
o
f
p
u
b
licly
av
ailab
le
b
en
ch
m
ar
k
d
atasets
o
f
th
is
k
in
d
,
r
ath
er
t
h
an
with
n
o
is
e
-
af
f
ec
te
d
,
ex
p
e
r
t
-
an
n
o
tated
f
ield
m
ea
s
u
r
em
e
n
ts
.
W
e
r
ep
o
r
t
th
is
r
esu
lt
as
ev
id
en
ce
th
at
th
e
p
r
o
p
o
s
ed
p
ip
elin
e
g
en
e
r
alize
s
s
u
cc
ess
f
u
lly
to
in
d
ep
en
d
en
tly
c
o
llected
d
ata
f
r
o
m
a
d
if
f
er
en
t
in
s
titu
tio
n
a
n
d
s
en
s
o
r
co
n
f
ig
u
r
atio
n
,
wh
ile
ac
k
n
o
w
led
g
in
g
th
at
s
tr
o
n
g
er
ev
id
e
n
ce
ag
ain
s
t
o
v
er
f
itti
n
g
wo
u
ld
r
e
q
u
ir
e
e
v
alu
atio
n
o
n
r
ea
l
-
tim
e,
co
n
tin
u
o
u
s
ly
lo
g
g
ed
f
ield
d
ata
with
n
atu
r
ally
o
cc
u
r
r
in
g
m
ea
s
u
r
em
en
t
n
o
is
e
—
a
d
ir
ec
tio
n
we
id
e
n
tif
y
ex
p
licitly
as a
p
r
i
o
r
ity
f
o
r
f
u
tu
r
e
wo
r
k
i
n
th
e
L
im
itatio
n
s
s
ec
tio
n
.
T
ab
le
9
.
E
x
ter
n
al
v
alid
atio
n
r
e
s
u
lts
o
n
an
in
d
ep
en
d
e
n
t d
ataset
[
2
8
]
,
n
=6
0
0
h
eld
-
o
u
t te
s
t sam
p
les
M
o
d
e
l
F1
-
sc
o
r
e
(
%)
R
e
c
a
l
l
(
%)
A
c
c
u
r
a
c
y
(
%)
M
LP
9
9
.
8
4
9
9
.
6
8
9
9
.
8
3
X
G
B
o
o
st
1
0
0
.
0
0
1
0
0
.
0
0
1
0
0
.
0
0
RF
1
0
0
.
0
0
1
0
0
.
0
0
1
0
0
.
0
0
S
t
a
c
k
i
n
g
1
0
0
.
0
0
1
0
0
.
0
0
1
0
0
.
0
0
4.
CO
NCLU
SI
O
N
T
h
is
s
tu
d
y
p
r
esen
ted
a
s
tack
in
g
en
s
em
b
le
co
m
b
in
in
g
RF
,
X
GB
o
o
s
t,
an
d
a
m
u
ltil
ay
er
p
er
ce
p
tr
o
n
f
o
r
b
in
ar
y
ir
r
ig
ati
o
n
p
u
m
p
-
s
tatu
s
p
r
ed
ictio
n
f
r
o
m
s
o
il
m
o
is
tu
r
e,
tem
p
er
atu
r
e,
h
u
m
id
ity
,
an
d
tim
e
-
of
-
d
a
y
s
en
s
o
r
r
ea
d
in
g
s
.
Usi
n
g
a
h
eld
-
o
u
t
t
est
s
et
with
5
-
f
o
ld
cr
o
s
s
-
v
alid
atio
n
,
th
e
s
tack
in
g
m
o
d
el
ac
h
iev
ed
9
9
.
9
7
%
ac
cu
r
ac
y
,
0
.
9
9
9
4
p
r
ec
is
io
n
,
1
.
0
0
0
0
r
ec
all,
0
.
9
9
9
4
s
p
ec
if
icity
,
an
d
0
.
9
9
9
7
F1
-
s
co
r
e,
m
a
r
g
in
ally
o
u
tp
er
f
o
r
m
in
g
XGBo
o
s
t
—
th
e
s
tr
o
n
g
est
in
d
iv
id
u
al
b
ase
lear
n
e
r
—
wh
ile
e
lim
in
atin
g
f
alse
n
eg
ativ
es.
Ho
wev
er
,
th
is
m
ar
g
in
is
s
m
all
r
elativ
e
to
cr
o
s
s
-
v
ali
d
atio
n
v
ar
ian
ce
a
n
d
s
h
o
u
l
d
n
o
t
b
e
o
v
er
-
in
ter
p
r
ete
d
.
An
aly
s
is
o
f
b
ase
-
lear
n
er
p
r
ed
ictio
n
ag
r
ee
m
en
t
i
d
en
tifie
d
r
esid
u
al
d
is
ag
r
ee
m
e
n
t
b
etw
ee
n
RF
,
XGBo
o
s
t
,
an
d
ML
P
as
th
e
m
ec
h
an
is
tic
s
o
u
r
ce
o
f
th
is
g
ain
.
A
d
ep
lo
y
m
en
t
-
o
r
ien
te
d
an
aly
s
is
s
h
o
wed
th
e
f
u
ll
en
s
em
b
le'
s
lar
g
er
m
em
o
r
y
f
o
o
tp
r
i
n
t
an
d
in
f
er
en
ce
co
s
t
s
u
it
g
atew
ay
-
c
lass
I
o
T
h
ar
d
war
e
b
etter
th
an
m
icr
o
co
n
tr
o
ller
-
class
d
ev
ices,
wh
ile
in
d
iv
id
u
al
XGBo
o
s
t
r
em
ain
s
v
iab
le
f
o
r
tig
h
ter
co
n
s
tr
ain
ts
.
SHAP
an
aly
s
is
id
en
tifie
d
tem
p
er
atu
r
e
an
d
tim
e
-
of
-
d
a
y
as
th
e
d
o
m
in
an
t,
p
h
y
s
ically
p
lau
s
ib
le
d
r
iv
er
s
o
f
p
r
ed
icted
p
u
m
p
ac
t
iv
atio
n
.
T
h
is
wo
r
k
h
as
lim
itati
o
n
s
.
T
h
e
d
ataset
is
d
er
iv
ed
f
r
o
m
a
p
u
b
licly
a
v
ailab
le
b
en
c
h
m
ar
k
r
at
h
er
t
h
an
c
o
n
tin
u
o
u
s
ly
l
o
g
g
e
d
f
ield
s
en
s
o
r
s
,
an
d
th
e
v
er
y
h
i
g
h
ac
cu
r
ac
y
—
co
r
r
o
b
o
r
ated
o
n
a
s
ec
o
n
d
,
in
d
e
p
en
d
e
n
t
d
ataset
—
lik
ely
r
ef
lects
a
n
ea
r
-
d
ete
r
m
in
is
tic
s
en
s
o
r
-
to
-
lab
el
r
elatio
n
s
h
ip
ch
ar
ac
ter
is
tic
o
f
s
u
ch
b
en
ch
m
ar
k
s
.
T
h
e
M
L
P
b
ase
lear
n
er
also
s
h
o
wed
co
n
s
is
ten
tly
wea
k
er
,
h
ig
h
er
-
v
ar
ian
ce
p
e
r
f
o
r
m
an
ce
th
an
th
e
tr
ee
-
b
ased
lear
n
er
s
.
Fu
tu
r
e
wo
r
k
s
h
o
u
ld
v
alid
ate
th
e
f
r
am
ewo
r
k
o
n
r
ea
l
-
tim
e
f
ield
d
ata
with
n
atu
r
ally
o
cc
u
r
r
i
n
g
s
en
s
o
r
n
o
is
e
an
d
ass
ess
o
n
-
d
ev
ice
co
m
p
u
tatio
n
al
f
ea
s
ib
ilit
y
.
ACK
NO
WL
E
DG
M
E
N
T
T
h
e
au
th
o
r
s
h
av
e
n
o
s
p
ec
if
ic
a
ck
n
o
wled
g
e
m
en
ts
to
d
ec
lar
e.
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
r
esear
ch
g
r
an
t
o
r
co
n
tr
ac
t
f
r
o
m
an
y
f
u
n
d
i
n
g
ag
en
c
y
in
th
e
p
u
b
lic,
co
m
m
er
cial,
o
r
n
o
t
-
f
o
r
-
p
r
o
f
it secto
r
s
.
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
t
r
ib
u
to
r
R
o
les
T
a
x
o
n
o
m
y
(
C
R
ed
iT
)
to
r
ec
o
g
n
ize
in
d
iv
i
d
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
Sar
r
a
Go
u
r
ar
i
✓
✓
✓
✓
✓
W
af
a
Dif
allah
✓
✓
B
elk
ac
em
Dr
ao
u
i
✓
✓
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
Evaluation Warning : The document was created with Spire.PDF for Python.