I
nte
rna
t
io
na
l J
o
urna
l o
f
P
o
wer
E
lect
ro
nics
a
nd
Driv
e
S
y
s
t
em
(
I
J
P
E
DS)
Vo
l.
1
7
,
No
.
3
,
Sep
tem
b
er
20
2
6
,
p
p
.
1
7
8
0
~
1
7
9
3
I
SS
N:
2
0
8
8
-
8
6
9
4
,
DOI
: 1
0
.
1
1
5
9
1
/ijp
ed
s
.
v
1
7
.
i
3
.
p
p
1
7
8
0
-
1
7
9
3
1780
J
o
ur
na
l
ho
m
ep
a
g
e
:
h
ttp
:
//ij
p
e
d
s
.
ia
esco
r
e.
co
m
Tra
nsfo
rmer f
a
ul
t
dia
g
no
sis
using
diss
o
lv
ed ga
s a
na
l
y
sis
:
a
hybrid ensem
bl
e mo
del wi
th
da
ta
preprocess
ing
F
a
t
ima
Z
o
hra
B
o
ud
j
ella
1
,
So
uh
ila
B
o
ud
j
el
la
2
,
Na
s
ir
u
Y
a
ha
y
a
Ahmed
3
,
H
a
zlee
Azil
I
lli
a
s
3
,
Sm
a
il L
a
t
if
a
4
,
Rer
iba
lla
h
H
a
f
id
h
a
5
1
En
g
i
n
e
e
r
i
n
g
a
n
d
S
u
s
t
a
i
n
a
b
l
e
D
e
v
e
l
o
p
me
n
t
La
b
o
r
a
t
o
r
y
,
D
e
p
a
r
t
me
n
t
o
f
El
e
c
t
r
i
c
a
l
E
n
g
i
n
e
e
r
i
n
g
,
U
n
i
v
e
r
si
t
y
o
f
A
i
n
Te
m
o
u
c
h
e
n
t
,
A
i
n
Te
m
o
u
c
h
e
n
t
,
A
l
g
e
r
i
a
2
LG
P
C
S
L
a
b
o
r
a
t
o
r
y
,
D
e
p
a
r
t
m
e
n
t
o
f
E
l
e
c
t
r
i
c
a
l
E
n
g
i
n
e
e
r
i
n
g
,
U
n
i
v
e
r
si
t
y
o
f
M
a
sca
r
a
,
M
a
s
c
a
r
a
,
A
l
g
e
r
i
a
3
D
e
p
a
r
t
me
n
t
o
f
El
e
c
t
r
i
c
a
l
En
g
i
n
e
e
r
i
n
g
,
F
a
c
u
l
t
y
o
f
E
n
g
i
n
e
e
r
i
n
g
,
U
n
i
v
e
r
si
t
i
M
a
l
a
y
a
,
K
u
a
l
a
L
u
m
p
u
r
,
M
a
l
a
y
s
i
a
4
LSTER
La
b
o
r
a
t
o
r
y
,
D
e
p
a
r
t
me
n
t
o
f
E
l
e
c
t
r
i
c
a
l
E
n
g
i
n
e
e
r
i
n
g
,
U
n
i
v
e
r
si
t
y
C
e
n
t
e
r
o
f
El
B
a
y
a
d
h
,
El
B
a
y
a
d
h
,
A
l
g
e
r
i
a
5
G
I
D
D
La
b
o
r
a
t
o
r
y
,
D
e
p
a
r
t
m
e
n
t
o
f
E
l
e
c
t
r
i
c
a
l
E
n
g
i
n
e
e
r
i
n
g
a
n
d
A
u
t
o
m
a
t
i
o
n
,
U
n
i
v
e
r
si
t
y
A
h
me
d
Z
a
b
a
n
a
o
f
R
e
l
i
z
a
n
e
,
R
e
l
i
z
a
n
e
,
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
Ma
r
1
3
,
2
0
2
6
R
ev
is
ed
J
u
l 1
4
,
2
0
2
6
Acc
ep
ted
J
u
l 2
1
,
2
0
2
6
F
a
il
u
re
s
a
n
d
g
u
a
ra
n
tee
d
d
e
p
e
n
d
a
b
il
it
y
o
f
t
h
e
e
lec
tri
c
a
l
g
ri
d
,
e
a
rly
fa
u
lt
d
iag
n
o
sis
i
n
p
o
we
r
tra
n
sfo
rm
e
rs
is
e
ss
e
n
ti
a
l.
B
y
e
x
a
m
in
i
n
g
g
a
s
ra
ti
o
s
su
g
g
e
sti
v
e
o
f
fa
u
lt
s,
d
isso
l
v
e
d
g
a
s
a
n
a
ly
sis
(D
G
A)
c
o
n
ti
n
u
e
s
to
b
e
a
v
it
a
l
c
o
m
p
o
n
e
n
t
f
o
r
tran
sfo
rm
e
r
h
e
a
lt
h
m
o
n
it
o
rin
g
.
Us
in
g
f
o
u
r
p
re
p
ro
c
e
ss
in
g
tec
h
n
iq
u
e
s
ra
w
d
a
ta,
m
in
-
m
a
x
n
o
rm
a
li
z
a
ti
o
n
,
l
o
g
a
rit
h
m
ic
tran
sfo
rm
a
ti
o
n
,
a
n
d
sq
u
a
re
ro
o
t
tran
sfo
rm
a
ti
o
n
;
th
is
stu
d
y
su
g
g
e
sts
a
m
a
c
h
in
e
lea
rn
in
g
m
e
th
o
d
fo
r
fa
u
lt
d
e
tec
ti
o
n
u
s
in
g
DG
A
g
a
s
ra
ti
o
s
(su
c
h
a
s
CH₄
/H₂
,
C₂
H₂
/C₂
H
₄
).
Ra
n
d
o
m
f
o
re
st
(RF
)
,
su
p
p
o
r
t
v
e
c
to
r
m
a
c
h
i
n
e
s
(S
VM)
,
g
ra
d
ien
t
b
o
o
ste
d
tree
s
(G
BT),
a
n
d
a
h
y
b
rid
RF
-
G
BT
m
o
d
e
l
th
a
t
u
se
s
p
re
d
ictio
n
fu
sio
n
we
re
th
e
f
o
u
r
su
p
e
rv
ise
d
c
las
sifiers
a
ss
e
ss
e
d
.
P
e
rfo
rm
a
n
c
e
wa
s
a
ss
e
ss
e
d
u
sin
g
a
c
c
u
ra
c
y
,
p
re
c
is
io
n
,
re
c
a
ll
,
f1
-
sc
o
re
,
a
n
d
C
o
h
e
n
'
s
k
a
p
p
a
.
Ex
p
e
rime
n
tal
re
su
lt
s
s
h
o
w
t
h
a
t
t
h
e
h
y
b
rid
R
F
-
G
BT
m
o
d
e
l
with
l
o
g
a
rit
h
m
ic
tran
sfo
rm
a
ti
o
n
a
c
h
iev
e
s
th
e
h
i
g
h
e
st
p
e
rfo
rm
a
n
c
e
,
with
9
4
.
9
3
%
a
c
c
u
ra
c
y
a
n
d
9
2
.
3
7
%
Co
h
e
n
'
s
k
a
p
p
a
,
sig
n
ifi
c
a
n
tl
y
o
u
tp
e
rf
o
rm
in
g
in
d
iv
i
d
u
a
l
c
las
sifiers
.
Da
ta p
re
p
ro
c
e
ss
in
g
,
p
a
rt
icu
larly
l
o
g
a
rit
h
m
ic an
d
sq
u
a
re
ro
o
t
tran
sf
o
rm
a
ti
o
n
s,
e
n
h
a
n
c
e
s
d
ia
g
n
o
stic
r
o
b
u
stn
e
ss
b
y
m
it
i
g
a
ti
n
g
fe
a
tu
re
sk
e
wn
e
ss
.
T
h
is
st
u
d
y
u
n
d
e
rsc
o
re
s
t
h
e
imp
o
rta
n
c
e
o
f
tai
lo
re
d
p
re
p
ro
c
e
ss
in
g
a
n
d
e
n
se
m
b
le
m
e
th
o
d
s
fo
r
re
li
a
b
le t
ra
n
sfo
rm
e
r
fa
u
lt
d
iag
n
o
sis.
K
ey
w
o
r
d
s
:
Data
p
r
ep
r
o
ce
s
s
in
g
Dis
s
o
lv
ed
g
as a
n
aly
s
is
Gas r
atio
m
eth
o
d
s
Hy
b
r
id
en
s
em
b
le
m
o
d
els
L
o
g
ar
ith
m
ic
tr
a
n
s
f
o
r
m
atio
n
Ma
ch
in
e
lear
n
in
g
class
if
ier
s
Sq
u
ar
e
r
o
o
t tr
an
s
f
o
r
m
atio
n
T
h
is i
s
a
n
o
p
e
n
a
c
c
e
ss
a
rticle
u
n
d
e
r th
e
CC B
Y
-
SA
li
c
e
n
se
.
C
o
r
r
e
s
p
o
nd
ing
A
uth
o
r
:
Fatim
a
Z
o
h
r
a
B
o
u
d
jella
E
n
g
in
ee
r
in
g
an
d
Su
s
tain
ab
le
Dev
elo
p
m
en
t L
a
b
o
r
ato
r
y
,
Dep
ar
tm
en
t o
f
E
lectr
ical
E
n
g
in
ee
r
in
g
Un
iv
er
s
ity
o
f
Ain
T
e
m
o
u
c
h
en
t
Ain
T
em
o
u
ch
e
n
t,
Alg
er
ia
E
m
ail:
f
atim
a.
b
o
u
d
jella@
u
n
iv
-
tem
o
u
ch
e
n
t.e
d
u
.
d
z
1.
I
NT
RO
D
UCT
I
O
N
Po
we
r
tr
a
n
s
f
o
r
m
er
s
a
r
e
v
i
tal
c
o
m
p
o
n
e
n
ts
o
f
e
lec
tr
ic
al
g
r
id
s
,
en
ab
li
n
g
e
f
f
ici
e
n
t
e
n
er
g
y
co
n
v
er
s
i
o
n
an
d
tr
a
n
s
f
e
r
a
cr
o
s
s
v
o
l
ta
g
e
l
e
v
els
.
T
h
eir
f
a
il
u
r
e
ca
n
l
ea
d
t
o
p
r
o
lo
n
g
e
d
o
u
t
ag
es,
s
i
g
n
i
f
i
ca
n
t
f
i
n
a
n
c
ial
l
o
s
s
es,
a
n
d
s
af
et
y
r
is
k
s
f
o
r
p
er
s
o
n
n
el
a
n
d
eq
u
i
p
m
e
n
t
[
1
]
,
[
2
]
.
F
o
r
p
r
e
d
i
cti
v
e
m
a
in
te
n
a
n
c
e
t
o
r
e
d
u
ce
t
h
es
e
r
is
k
s
,
ef
f
e
cti
v
e
co
n
d
iti
o
n
m
o
n
it
o
r
i
n
g
is
t
h
e
r
e
f
o
r
e
cr
u
c
ial
.
B
y
e
x
a
m
i
n
i
n
g
g
as
es
d
is
s
o
l
v
e
d
i
n
in
s
u
l
ati
n
g
o
il,
d
is
s
o
l
v
e
d
g
as
an
al
y
s
is
(
DGA
)
,
a
c
o
m
m
o
n
ly
u
s
e
d
n
o
n
-
i
n
v
asi
v
e
m
et
h
o
d
,
ca
n
id
en
ti
f
y
t
r
a
n
s
f
o
r
m
er
f
a
u
lts
li
k
e
p
a
r
tia
l
d
is
c
h
ar
g
es
,
th
e
r
m
al
o
v
e
r
h
e
ati
n
g
,
a
n
d
el
e
ctr
ica
l
a
r
c
in
g
[
3
]
,
[
4
]
.
T
h
e
Du
v
al
T
r
ia
n
g
l
e,
R
o
g
e
r
s
R
at
i
o
s
,
a
n
d
D
o
r
n
e
n
b
u
r
g
ap
p
r
o
ac
h
a
r
e
e
x
a
m
p
les
o
f
t
r
a
d
i
tio
n
a
l
DGA
i
n
te
r
p
r
et
ati
o
n
t
ec
h
n
i
q
u
es
t
h
at
u
s
e
e
m
p
ir
ica
l
th
r
es
h
o
l
d
s
t
o
d
et
er
m
i
n
e
f
a
u
lt
t
y
p
es
[
5
]
.
H
o
we
v
er
,
t
h
e
s
e
te
ch
n
i
q
u
es
f
r
e
q
u
e
n
tl
y
p
r
o
d
u
c
e
i
n
c
o
n
s
is
t
en
t
d
ia
g
n
o
s
es
d
u
e
to
t
h
ei
r
l
ac
k
o
f
p
r
ec
is
i
o
n
in
a
m
b
ig
u
o
u
s
c
ases
o
r
n
o
n
-
s
ta
n
d
a
r
d
c
o
n
d
iti
o
n
s
.
New
o
p
p
o
r
t
u
n
iti
es
f
o
r
i
m
p
r
o
v
in
g
DG
A
r
eli
a
b
ili
ty
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J Po
w
E
lec
&
Dr
i Sy
s
t
I
SS
N:
2088
-
8
6
9
4
Tr
a
n
s
fo
r
mer fa
u
lt d
ia
g
n
o
s
is
u
s
in
g
d
is
s
o
lved
g
a
s
a
n
a
lysi
s
:
a
h
yb
r
id
en
s
emb
le
…
(
F
a
tima
Z
o
h
r
a
B
o
u
d
jella
)
1781
h
a
v
e
b
e
en
m
a
d
e
p
o
s
s
ib
le
b
y
t
h
e
d
e
v
e
lo
p
m
e
n
t
o
f
a
r
t
if
ici
al
i
n
te
lli
g
e
n
c
e
(
A
I
)
an
d
m
ac
h
i
n
e
l
ea
r
n
i
n
g
(
M
L
)
[
6
]
,
[
7
]
.
Nu
m
e
r
o
u
s
s
t
u
d
ies
h
a
v
e
i
n
v
e
s
tig
ate
d
t
h
e
a
p
p
li
ca
ti
o
n
o
f
i
n
te
lli
g
e
n
t
s
y
s
te
m
s
t
o
i
m
p
r
o
v
e
t
r
a
n
s
f
o
r
m
e
r
f
a
u
l
t
d
ia
g
n
o
s
is
.
Fo
r
ex
am
p
l
e,
a
m
u
l
t
is
ta
g
e
s
u
p
p
o
r
t
v
e
ct
o
r
m
ac
h
i
n
e
(
SVM
)
t
r
a
in
e
d
o
n
I
E
C
T
C
1
0
DGA
d
at
a
a
ch
ie
v
e
d
a
r
ec
o
g
n
i
ti
o
n
r
ate
o
f
a
b
o
u
t
9
0
%,
d
em
o
n
s
t
r
ati
n
g
t
h
e
p
o
t
en
t
ial
o
f
m
a
c
h
i
n
e
l
ea
r
n
i
n
g
tec
h
n
iq
u
es
f
o
r
ac
cu
r
a
te
tr
a
n
s
f
o
r
m
e
r
f
au
lt c
l
ass
if
ic
ati
o
n
[
8
]
.
T
h
e
co
m
b
i
n
at
io
n
o
f
m
ac
h
in
e
l
ea
r
n
i
n
g
m
et
h
o
d
s
an
d
c
o
n
v
en
t
io
n
al
d
ia
g
n
o
s
t
ic
in
d
ic
at
o
r
s
is
s
till
b
e
in
g
i
n
v
esti
g
at
e
d
in
r
e
ce
n
t
r
ese
ar
c
h
.
Ni
țu
et
a
l
.
[
9
]
p
r
o
p
o
s
ed
a
m
e
th
o
d
c
o
m
b
i
n
i
n
g
t
h
e
t
h
r
ee
r
at
io
te
ch
n
i
q
u
e
wit
h
a
r
a
n
d
o
m
f
o
r
es
t
(
R
F
)
class
if
ie
r
t
o
i
d
e
n
ti
f
y
t
r
a
n
s
f
o
r
m
e
r
f
a
u
l
ts
u
s
i
n
g
DGA
g
as
r
at
io
in
d
ic
at
o
r
s
.
A
n
o
t
h
e
r
s
t
u
d
y
p
r
o
p
o
s
ed
a
d
ia
g
n
o
s
t
ic
f
r
am
ew
o
r
k
b
as
ed
o
n
a
d
v
a
n
c
ed
f
e
at
u
r
e
e
x
t
r
a
cti
o
n
a
n
d
a
L
i
g
h
tGB
M
e
n
s
e
m
b
le
m
o
d
e
l,
wh
e
r
e
d
a
ta
p
r
ep
r
o
c
ess
i
n
g
a
n
d
f
e
at
u
r
e
s
ele
cti
o
n
im
p
r
o
v
e
d
cla
s
s
if
i
ca
t
io
n
a
cc
u
r
ac
y
f
r
o
m
8
6
.
8
4
%
to
9
3
.
4
2
%
[
1
0
]
.
N
o
twi
th
s
ta
n
d
in
g
t
h
es
e
d
ev
el
o
p
m
e
n
ts
,
DGA
-
b
ase
d
d
i
ag
n
o
s
is
s
till
f
a
ce
s
d
i
f
f
ic
u
lt
ies
.
C
lass
im
b
a
la
n
c
e,
e
x
t
r
e
m
e
v
a
lu
es,
a
n
d
o
p
er
ati
o
n
al
v
a
r
i
ab
ilit
y
ar
e
e
x
am
p
l
es o
f
d
a
t
a
q
u
a
lit
y
p
r
o
b
l
em
s
th
a
t
c
a
n
i
m
p
ai
r
m
o
d
el
p
e
r
f
o
r
m
an
ce
[
1
1
]
.
F
u
r
th
er
m
o
r
e
,
s
ta
n
d
a
r
d
s
s
u
c
h
as I
E
C
6
0
5
9
9
r
ec
o
m
m
e
n
d
t
h
e
u
s
e
o
f
g
as
r
at
io
s
f
o
r
m
o
r
e
r
eli
ab
le
f
a
u
l
t d
i
ag
n
o
s
is
[
1
2
]
.
I
n
a
d
d
iti
o
n
,
th
e
c
h
o
ic
e
o
f
d
ata
p
r
e
p
r
o
ce
s
s
in
g
te
ch
n
i
q
u
es
,
i
n
c
lu
d
i
n
g
n
o
r
m
al
iz
ati
o
n
a
n
d
d
a
ta
t
r
a
n
s
f
o
r
m
at
io
n
,
ca
n
s
i
g
n
i
f
ic
a
n
tl
y
i
n
f
lu
e
n
ce
class
if
ie
r
p
er
f
o
r
m
a
n
ce
,
y
e
t
t
h
is
as
p
e
ct
r
e
m
ai
n
s
i
n
s
u
f
f
ic
ie
n
tl
y
i
n
v
esti
g
ate
d
in
t
h
e
l
ite
r
a
tu
r
e
.
T
o
ad
d
r
ess
th
ese
lim
itatio
n
s
,
th
is
s
tu
d
y
g
o
es
b
ey
o
n
d
co
n
v
en
tio
n
al
ap
p
r
o
ac
h
es
b
y
ex
p
licitly
in
v
esti
g
atin
g
th
e
i
n
ter
ac
tio
n
b
etwe
en
d
ata
p
r
ep
r
o
ce
s
s
in
g
tech
n
iq
u
es
an
d
e
n
s
em
b
le
lear
n
i
n
g
m
o
d
els
f
o
r
DGA
-
b
ased
tr
an
s
f
o
r
m
er
f
au
lt
d
iag
n
o
s
is
.
W
h
ile
p
r
ev
io
u
s
wo
r
k
s
ty
p
ically
f
o
cu
s
o
n
eith
e
r
class
if
ier
o
p
tim
izatio
n
o
r
s
in
g
le
p
r
ep
r
o
ce
s
s
in
g
s
tr
ateg
ies,
th
e
co
m
b
in
e
d
ef
f
ec
t
o
f
m
u
ltip
le
d
ata
tr
an
s
f
o
r
m
atio
n
s
o
n
en
s
em
b
le
m
o
d
el
p
er
f
o
r
m
an
ce
r
e
m
ain
s
in
s
u
f
f
ici
en
tly
ex
p
lo
r
ed
.
I
n
th
is
co
n
te
x
t,
g
as
r
atio
s
ar
e
em
p
lo
y
ed
as
d
iag
n
o
s
tic
f
ea
tu
r
es
in
s
tead
o
f
r
aw
g
as
co
n
ce
n
t
r
atio
n
s
to
en
s
u
r
e
r
o
b
u
s
tn
ess
an
d
alig
n
m
en
t
with
I
E
C
s
tan
d
ar
d
s
.
M
u
ltip
le
p
r
ep
r
o
ce
s
s
in
g
tech
n
i
q
u
es,
in
clu
d
in
g
r
aw
d
ata,
m
in
–
m
a
x
n
o
r
m
aliza
tio
n
,
lo
g
ar
ith
m
ic
tr
an
s
f
o
r
m
atio
n
,
a
n
d
s
q
u
ar
e
-
r
o
o
t
tr
an
s
f
o
r
m
atio
n
,
a
r
e
s
y
s
tem
atica
lly
ev
alu
ated
to
ass
es
s
th
eir
in
f
lu
en
ce
o
n
f
ea
t
u
r
e
d
is
tr
ib
u
tio
n
an
d
class
if
icatio
n
p
er
f
o
r
m
an
ce
.
Fu
r
th
er
m
o
r
e
,
th
is
s
tu
d
y
in
tr
o
d
u
c
es
a
h
y
b
r
id
R
F
–
GB
T
m
o
d
el
b
ased
o
n
p
r
e
d
ictio
n
f
u
s
io
n
,
d
esig
n
ed
to
ex
p
lo
it
th
e
co
m
p
lem
en
tar
y
s
tr
en
g
t
h
s
o
f
en
s
em
b
le
lear
n
in
g
tech
n
iq
u
es
.
Un
lik
e
tr
ad
itio
n
al
b
en
ch
m
ar
k
in
g
s
tu
d
ies,
th
e
p
r
o
p
o
s
ed
ap
p
r
o
ac
h
em
p
h
asizes
th
e
jo
in
t
o
p
tim
izatio
n
o
f
f
ea
tu
r
e
r
ep
r
esen
tatio
n
an
d
m
o
d
el
ar
ch
itectu
r
e,
p
r
o
v
id
i
n
g
a
s
tr
u
ctu
r
ed
f
r
am
ewo
r
k
f
o
r
im
p
r
o
v
i
n
g
d
iag
n
o
s
tic
r
eliab
ilit
y
.
T
h
e
m
o
d
els
ar
e
ev
alu
ated
th
r
o
u
g
h
m
u
lti
-
class
f
au
lt
class
if
icatio
n
(
elec
tr
ical
f
au
lts
,
th
er
m
al
f
au
lts
,
an
d
n
o
r
m
al
co
n
d
itio
n
s
)
u
s
in
g
co
m
p
r
eh
en
s
iv
e
p
er
f
o
r
m
an
ce
m
etr
ics
s
u
ch
as
ac
cu
r
ac
y
,
p
r
ec
is
io
n
,
r
ec
all,
F1
-
s
co
r
e,
a
n
d
C
o
h
en
’
s
k
ap
p
a
.
Ov
er
all,
th
e
m
ain
co
n
tr
i
b
u
tio
n
o
f
th
is
wo
r
k
lies
in
p
r
o
v
i
d
in
g
a
s
y
s
tem
atic
an
d
ap
p
licatio
n
-
o
r
ien
ted
f
r
a
m
ewo
r
k
th
at
b
r
id
g
es
p
r
ep
r
o
ce
s
s
in
g
s
tr
ateg
ies
an
d
en
s
em
b
le
lear
n
in
g
,
th
er
eb
y
o
f
f
e
r
in
g
im
p
r
o
v
ed
r
o
b
u
s
tn
ess
an
d
r
eliab
ilit
y
in
DGA
-
b
ased
tr
an
s
f
o
r
m
er
f
au
lt d
iag
n
o
s
is
.
2.
M
E
T
H
O
D
2
.
1
.
Da
t
a
s
et
T
h
e
d
ataset
was
d
er
i
v
ed
f
r
o
m
DGA
r
ep
o
r
ts
o
f
o
p
e
r
atio
n
al
p
o
wer
tr
an
s
f
o
r
m
er
s
lo
ca
te
d
in
v
ar
io
u
s
p
r
o
v
in
ce
s
ac
r
o
s
s
Alg
er
ia,
co
m
p
r
is
in
g
6
4
8
s
am
p
les.
I
n
s
tead
o
f
ab
s
o
lu
te
g
as
co
n
ce
n
tr
ati
o
n
s
(
p
p
m
)
,
g
as
r
atio
s
(
C
H₄/H₂
,
C
₂H₂
/
C
₂H₄
,
C
₂H₆
/
C
H₄,
C
₂H₄
/C
₂H₆
)
wer
e
u
s
ed
,
a
s
r
ec
o
m
m
en
d
ed
b
y
t
h
e
Mo
d
if
ied
R
o
g
er
s
’
R
atio
Me
th
o
d
[
1
3
]
,
to
en
h
an
ce
r
o
b
u
s
tn
ess
ag
ain
s
t
o
p
er
atio
n
al
v
ar
iab
ilit
y
.
T
h
e
s
am
p
les
wer
e
ev
en
ly
d
is
tr
ib
u
ted
ac
r
o
s
s
th
r
ee
f
au
lt c
lass
es:
i)
E
lectr
ical
f
au
lts
(
E
)
:
p
ar
tial d
i
s
ch
ar
g
es a
n
d
ar
cin
g
(
2
1
6
s
am
p
les).
ii)
T
h
er
m
al
f
au
lts
(
T
)
:
c
ellu
lo
s
e
o
r
co
n
d
u
cto
r
o
v
er
h
ea
tin
g
(
2
1
6
s
am
p
les).
iii)
No
r
m
al
s
tate
(
N)
:
n
o
f
a
u
lt d
et
ec
ted
(
2
1
6
s
am
p
les).
E
ac
h
s
am
p
le
in
clu
d
es
f
o
u
r
g
as
r
atio
s
s
elec
ted
f
o
r
th
eir
d
iag
n
o
s
tic
r
elev
a
n
ce
.
T
a
b
le
1
p
r
esen
ts
r
ep
r
esen
tativ
e
ex
am
p
les
o
f
d
i
s
s
o
lv
ed
g
as
co
n
ce
n
t
r
atio
n
s
f
o
r
ea
ch
tr
a
n
s
f
o
r
m
er
co
n
d
itio
n
(
elec
tr
ical,
th
er
m
al,
an
d
n
o
r
m
al)
,
illu
s
tr
atin
g
th
e
ty
p
ical
d
iag
n
o
s
tic
p
atter
n
s
o
b
s
er
v
ed
f
o
r
ea
ch
ty
p
e
o
f
f
au
lt.
T
h
ese
f
o
u
r
r
atio
s
C
H₄/H₂
,
C
₂H₂
/
C
₂H₄
,
C
₂H₆
/C
H₄
,
an
d
C
₂H₄
/C
₂H₆
wer
e
s
elec
ted
b
ased
o
n
th
eir
d
iag
n
o
s
tic
r
elev
an
ce
,
as
th
ey
ar
e
co
m
m
o
n
l
y
u
s
ed
in
i
n
d
u
s
tr
y
s
ta
n
d
ar
d
s
an
d
r
ef
lect
d
is
tin
ct
f
au
lt m
ec
h
an
is
m
s
.
T
ab
le
1
.
R
ep
r
esen
tativ
e
g
as c
o
n
ce
n
tr
atio
n
s
(
p
p
m
)
f
o
r
ea
ch
tr
a
n
s
f
o
r
m
er
c
o
n
d
itio
n
H2
C
H
4
C
2
H
2
C
2
H
4
C
2
H
6
S
t
a
t
e
n
u
m
b
e
r
5
2
3
.
00
44
.
00
1
.
50
1
.
50
17
.
00
E
2
1
6
2
5
2
6
.
30
1
3
0
.
55
1
.
00
1
.
53
15
.
00
2
9
1
1
2
6
0
2
3
2
6
9
T
2
1
6
3
9
1
1
4
2
9
1
6
2
7
6
1
2
3
1
0
.
1
1
.
1
2
.
02
3
.
1
4
.
02
N
2
1
6
2
2
.
5
0
.
5
0
5
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8
6
9
4
I
n
t J Po
w
E
lec
&
Dr
i Sy
s
t
,
Vo
l.
1
7
,
No
.
3
,
Sep
tem
b
er
20
2
6
:
178
0
-
1
7
9
3
1782
DGA
d
ata
f
o
r
t
r
an
s
f
o
r
m
e
r
d
ia
g
n
o
s
tics
o
f
ten
ex
h
ib
it n
o
n
-
lin
e
ar
d
is
tr
ib
u
tio
n
s
,
v
ar
y
in
g
s
ca
les,
an
d
h
ig
h
v
ar
iab
ilit
y
.
Sev
er
al
f
ac
to
r
s
co
n
tr
ib
u
te
to
th
is
v
ar
iab
ilit
y
,
s
u
ch
as
o
p
er
atin
g
co
n
d
itio
n
s
,
f
au
lt
d
u
r
atio
n
,
m
ea
s
u
r
em
en
t
in
ac
cu
r
ac
ies,
an
d
en
v
ir
o
n
m
en
tal
in
f
lu
e
n
ce
s
.
Dir
ec
t
p
r
o
ce
s
s
in
g
o
f
th
ese
r
aw
d
ata
ca
n
lead
to
a
s
ig
n
if
ican
t
d
eg
r
ad
atio
n
in
t
h
e
p
er
f
o
r
m
an
ce
o
f
m
ac
h
in
e
lear
n
in
g
m
o
d
els,
esp
ec
ially
in
alg
o
r
ith
m
s
s
en
s
itiv
e
to
s
ca
les.
Prio
r
to
tr
an
s
f
o
r
m
ati
o
n
,
all
s
am
p
les
co
n
tain
i
n
g
m
is
s
in
g
v
alu
es
wer
e
r
em
o
v
e
d
to
en
s
u
r
e
d
ataset
co
m
p
leten
ess
an
d
to
av
o
id
in
t
r
o
d
u
cin
g
b
ias
d
u
r
in
g
m
o
d
el
tr
ain
in
g
.
Data
p
r
ep
r
o
ce
s
s
in
g
an
d
n
o
r
m
aliza
tio
n
aim
to
m
itig
ate
is
s
u
es
r
elate
d
to
s
ca
le,
d
is
tr
ib
u
tio
n
s
k
ewn
ess
,
an
d
n
o
is
e.
T
h
is
s
tep
is
p
ar
ticu
lar
ly
im
p
o
r
tan
t
f
o
r
alg
o
r
ith
m
s
s
en
s
itiv
e
to
in
p
u
t
d
is
tr
ib
u
tio
n
s
,
s
u
ch
as SVM
:
i)
R
ed
u
ce
th
e
im
p
ac
t
o
f
lar
g
e
a
m
p
litu
d
es:
So
m
e
v
ar
ia
b
les,
s
u
ch
as
h
y
d
r
o
g
en
o
r
eth
y
le
n
e,
ca
n
r
ea
ch
v
er
y
h
ig
h
v
al
u
es c
o
m
p
ar
e
d
to
o
th
er
s
,
th
er
eb
y
s
k
ewin
g
d
is
tan
ce
ca
lcu
latio
n
s
o
r
g
r
a
d
ien
ts
.
ii)
E
n
s
u
r
e
f
air
n
ess
b
etwe
en
v
ar
iab
les:
B
y
s
ca
lin
g
all
v
ar
iab
les
t
o
a
co
m
m
o
n
r
an
g
e,
we
p
r
ev
en
t
an
y
s
in
g
le
g
as
f
r
o
m
d
o
m
in
atin
g
th
e
m
o
d
el’
s
d
ec
is
io
n
.
iii)
I
m
p
r
o
v
e
alg
o
r
ith
m
c
o
n
v
er
g
e
n
ce
:
Fo
r
m
o
d
els
lik
e
SVM,
n
o
r
m
aliza
tio
n
e
n
ab
les
f
aster
a
n
d
m
o
r
e
s
tab
le
co
n
v
er
g
en
ce
.
iv
)
R
ed
u
ce
th
e
ef
f
ec
t
o
f
e
x
tr
em
e
o
r
o
u
tlier
v
alu
es:
C
er
tain
tr
an
s
f
o
r
m
atio
n
s
,
s
u
ch
as
lo
g
ar
i
th
m
s
o
r
s
q
u
ar
e
r
o
o
ts
,
co
m
p
r
ess
h
ig
h
v
al
u
es,
m
ak
in
g
d
is
tr
ib
u
tio
n
s
m
o
r
e
s
y
m
m
etr
ic.
A
l
t
h
o
u
g
h
t
h
e
d
a
t
a
s
e
t
i
s
d
e
r
i
v
e
d
f
r
o
m
o
p
e
r
a
t
i
o
n
a
l
t
r
a
n
s
f
o
r
m
e
r
s
l
o
c
a
t
e
d
i
n
v
a
r
i
o
u
s
p
r
o
v
i
n
c
e
s
a
c
r
o
s
s
A
l
g
e
r
i
a
,
i
t
i
n
cl
u
d
e
s
a
d
i
v
e
r
s
e
r
a
n
g
e
o
f
o
p
e
r
a
t
i
n
g
c
o
n
d
i
t
i
o
n
s
a
n
d
f
a
u
l
t
s
c
e
n
a
r
i
o
s
r
e
p
r
e
s
e
n
t
a
t
i
v
e
o
f
r
e
a
l
-
w
o
r
l
d
p
o
w
e
r
s
y
s
t
e
m
s
.
T
h
e
d
a
t
as
e
t
co
v
e
r
s
t
h
r
e
e
m
a
j
o
r
c
l
a
s
s
es
o
f
t
r
a
n
s
f
o
r
m
e
r
c
o
n
d
i
t
i
o
n
s
(
e
l
ec
t
r
ic
a
l
f
a
u
l
ts
,
t
h
e
r
m
a
l
f
a
u
l
t
s
,
a
n
d
n
o
r
m
a
l
o
p
e
r
a
t
i
o
n
)
,
w
h
i
c
h
c
o
r
r
e
s
p
o
n
d
t
o
t
h
e
m
o
s
t
c
o
m
m
o
n
f
a
u
l
t
c
a
t
e
g
o
r
i
e
s
r
e
p
o
r
t
e
d
i
n
t
h
e
l
i
t
e
r
a
t
u
r
e
a
n
d
i
n
d
u
s
t
r
i
al
s
t
a
n
d
a
r
d
s
.
T
h
e
u
s
e
o
f
g
a
s
r
a
ti
o
s
i
n
s
t
e
a
d
o
f
a
b
s
o
l
u
t
e
g
a
s
c
o
n
c
e
n
t
r
a
ti
o
n
s
f
u
r
t
h
e
r
e
n
h
a
n
c
e
s
t
h
e
r
o
b
u
s
t
n
e
s
s
a
n
d
g
e
n
e
r
a
l
i
z
a
b
i
l
ity
o
f
t
h
e
p
r
o
p
o
s
e
d
a
p
p
r
o
a
c
h
.
U
n
l
i
k
e
p
p
m
v
a
l
u
e
s
,
g
a
s
r
a
t
i
o
s
a
r
e
l
e
s
s
s
e
n
s
i
ti
v
e
t
o
t
r
a
n
s
f
o
r
m
e
r
s
iz
e
,
o
i
l
v
o
l
u
m
e,
a
n
d
o
p
e
r
a
t
i
o
n
a
l
c
o
n
d
i
t
i
o
n
s
,
m
a
k
i
n
g
t
h
e
m
m
o
r
e
s
u
i
t
a
b
le
f
o
r
c
r
o
s
s
-
s
y
s
t
em
a
p
p
l
i
c
a
t
i
o
n
s
.
T
h
is
a
li
g
n
s
wi
t
h
I
E
C
6
0
5
9
9
r
e
c
o
m
m
e
n
d
a
t
i
o
n
s
,
w
h
e
r
e
r
a
t
i
o
-
b
as
e
d
i
n
t
e
r
p
r
et
a
ti
o
n
i
s
p
r
e
f
e
r
r
e
d
f
o
r
c
o
n
s
i
s
t
e
n
t
f
a
u
lt
d
i
a
g
n
o
s
is
.
H
o
w
e
v
e
r
,
i
t
i
s
ac
k
n
o
w
l
e
d
g
e
d
t
h
a
t
t
h
e
d
a
t
as
e
t
o
r
i
g
i
n
at
e
s
f
r
o
m
a
s
p
e
c
i
f
i
c
g
e
o
g
r
a
p
h
i
c
al
a
n
d
o
p
e
r
a
t
i
o
n
a
l
c
o
n
t
e
x
t
,
w
h
i
ch
m
a
y
l
i
m
i
t
i
ts
f
u
l
l
g
e
n
e
r
a
l
iza
t
i
o
n
.
F
u
t
u
r
e
w
o
r
k
w
i
ll
f
o
c
u
s
o
n
v
a
l
i
d
a
t
i
n
g
t
h
e
p
r
o
p
o
s
e
d
m
e
t
h
o
d
o
l
o
g
y
u
s
i
n
g
l
a
r
g
e
r
,
m
u
l
t
i
-
s
o
u
r
c
e
d
a
t
as
e
ts
f
r
o
m
d
i
f
f
e
r
e
n
t
r
e
g
i
o
n
s
a
n
d
t
r
an
s
f
o
r
m
e
r
t
y
p
e
s
t
o
f
u
r
t
h
e
r
a
s
s
ess
i
ts
s
c
a
l
a
b
il
i
t
y
a
n
d
r
o
b
u
s
t
n
e
s
s
.
I
n
t
h
i
s
w
o
r
k
,
f
o
u
r
m
a
i
n
d
a
t
a
t
r
a
n
s
f
o
r
m
a
t
i
o
n
m
e
t
h
o
d
s
w
e
r
e
u
s
e
d
.
2
.
1
.
1
.
Ra
w
da
t
a
Fig
u
r
e
1
illu
s
tr
ates
th
e
d
is
tr
ib
u
tio
n
o
f
th
e
o
r
i
g
in
al
(
n
o
n
-
tr
a
n
s
f
o
r
m
ed
)
d
ataset
u
s
ed
as
a
b
aselin
e
to
ev
alu
ate
th
e
ac
tu
al
ef
f
ec
t
o
f
p
r
ep
r
o
ce
s
s
in
g
tech
n
iq
u
es.
I
n
Fig
u
r
e
1
(
a
)
,
th
e
b
ar
ch
a
r
t
s
h
o
ws
th
e
v
ar
iatio
n
o
f
k
e
y
g
as
r
atio
s
ac
r
o
s
s
d
if
f
er
en
t
f
au
l
t
ty
p
es
(
elec
tr
ical
f
au
lts
(
E
)
,
th
er
m
al
f
au
lts
(
T
)
,
an
d
n
o
r
m
al
c
o
n
d
itio
n
(
N)
)
.
T
h
e
C
H₄/H₂
r
atio
is
clea
r
ly
d
o
m
in
an
t
f
o
r
th
er
m
al
f
a
u
lts
(
T
)
,
in
d
icatin
g
s
ig
n
if
ican
t
m
eth
an
e
p
r
o
d
u
ctio
n
d
u
e
to
o
il
o
v
er
h
ea
tin
g
,
m
ak
in
g
it
an
ex
ce
llen
t
in
d
icato
r
o
f
th
er
m
al
f
au
lts
.
T
h
e
C
₂H₄
/
C
₂H₆
r
atio
i
s
al
s
o
h
ig
h
f
o
r
th
er
m
a
l
f
au
lts
,
r
ef
lectin
g
m
o
d
er
ate
s
y
s
tem
o
v
er
h
ea
tin
g
,
wh
ile
it
r
em
ain
s
r
elativ
ely
lo
w
f
o
r
elec
tr
ic
al
f
au
lts
(
E
)
an
d
th
e
n
o
r
m
al
s
tate
(
N)
.
Fig
u
r
e
1
(
b
)
p
r
esen
ts
a
s
ca
tter
p
lo
t
o
f
C
₂
H₂/C
₂H₄
v
er
s
u
s
C
₂H₆
/
C
H₄
r
ati
o
s
.
T
h
e
d
is
tr
ib
u
tio
n
r
ev
ea
ls
a
h
ig
h
d
e
g
r
ee
o
f
o
v
er
lap
b
etwe
en
f
au
lt
class
es,
with
n
o
clea
r
lin
ea
r
o
r
s
ep
ar
a
b
le
b
o
u
n
d
ar
y
b
etwe
en
elec
tr
ical,
th
er
m
al,
an
d
n
o
r
m
a
l
co
n
d
itio
n
s
.
T
h
is
d
is
p
er
s
io
n
i
n
d
icate
s
s
tr
o
n
g
v
a
r
iab
ilit
y
an
d
n
o
n
lin
ea
r
ity
in
th
e
r
aw
d
ata,
wh
ich
m
a
k
es d
ir
ec
t
class
if
icatio
n
ch
allen
g
in
g
f
o
r
s
tan
d
ar
d
m
ac
h
in
e
lear
n
in
g
m
o
d
els.
(
a)
(
b
)
Fig
u
r
e
1.
Data
s
et
d
is
tr
ib
u
tio
n
(
r
aw
d
ata)
: (
a)
av
er
a
g
e
g
as r
ati
o
v
alu
es f
o
r
ea
ch
tr
a
n
s
f
o
r
m
e
r
f
au
lt c
ateg
o
r
y
an
d
(
b
)
s
ca
tter
p
lo
t illu
s
tr
atin
g
th
e
r
elatio
n
s
h
ip
b
etwe
en
th
e
C
₂H₂
/C
₂H₄
an
d
C
₂H
6
/C
H₄
g
as r
atio
s
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J Po
w
E
lec
&
Dr
i Sy
s
t
I
SS
N:
2088
-
8
6
9
4
Tr
a
n
s
fo
r
mer fa
u
lt d
ia
g
n
o
s
is
u
s
in
g
d
is
s
o
lved
g
a
s
a
n
a
lysi
s
:
a
h
yb
r
id
en
s
emb
le
…
(
F
a
tima
Z
o
h
r
a
B
o
u
d
jella
)
1783
2
.
1
.
2
.
L
o
g
a
rit
hm
ic
t
ra
ns
f
o
r
m
a
t
io
n
Usef
u
l f
o
r
lin
ea
r
izin
g
th
e
d
ata
an
d
r
ed
u
cin
g
t
h
e
in
f
lu
e
n
ce
o
f
ex
tr
em
e
v
alu
es.
′
=
l
og
(
+
1
)
(
1
)
L
o
g
ar
ith
m
ic
tr
an
s
f
o
r
m
atio
n
s
ig
n
if
ican
tly
en
h
a
n
ce
s
d
ata
d
is
tr
ib
u
tio
n
an
d
r
ea
d
ab
ilit
y
,
as
s
h
o
wn
in
Fig
u
r
e
2
.
I
n
th
e
b
ar
ch
a
r
t (
Fig
u
r
e
2
(
a
)
)
,
th
e
C
₂H₄
/C
₂
H₆
r
atio
is
elev
ated
f
o
r
th
er
m
al
f
a
u
lts
(
T
)
,
r
ef
lectin
g
o
v
er
h
ea
tin
g
,
wh
ile
th
e
C
₂H₂
/C
₂H₄
r
atio
is
h
ig
h
er
f
o
r
elec
tr
ical
f
au
lts
(
E
)
,
in
d
ic
ativ
e
o
f
ar
cin
g
o
r
p
ar
tial
d
is
c
h
ar
g
es.
T
h
e
C
H₄/H₂
r
atio
ef
f
ec
tiv
ely
d
is
cr
im
in
ates
th
er
m
al
f
au
lts
,
an
d
C
₂H₆
/C
H₄
i
s
p
r
o
m
in
en
t f
o
r
elec
tr
ical
f
au
lt
s
,
with
th
e
n
o
r
m
al
s
tate
(
N)
s
h
o
win
g
b
alan
ce
d
r
a
tio
s
.
T
h
e
s
ca
tter
p
lo
t
(
Fig
u
r
e
2
(
b
)
)
(
C
₂H₂
/C
₂H₄
v
s
.
C
₂H₆
/
C
H₄
)
r
ev
ea
ls
im
p
r
o
v
e
d
class
s
ep
ar
atio
n
f
o
r
E
,
T
,
a
n
d
N,
as
lo
g
ar
ith
m
ic
tr
an
s
f
o
r
m
atio
n
r
ed
u
ce
s
th
e
im
p
ac
t
o
f
ex
tr
em
e
v
al
u
es,
r
esu
ltin
g
in
m
o
r
e
u
n
if
o
r
m
d
is
tr
ib
u
tio
n
s
.
T
h
is
p
r
e
p
r
o
ce
s
s
in
g
en
h
an
ce
s
th
e
p
er
f
o
r
m
a
n
c
e
o
f
class
if
icatio
n
m
o
d
els.
(
a)
(
b
)
Fig
u
r
e
2
.
Data
s
et
d
is
tr
ib
u
tio
n
(
L
OG)
: (
a)
av
er
ag
e
g
as r
atio
v
alu
es f
o
r
ea
ch
t
r
an
s
f
o
r
m
e
r
f
au
l
t c
ateg
o
r
y
,
a
n
d
(
b
)
s
ca
tter
p
lo
t illu
s
tr
atin
g
th
e
r
elatio
n
s
h
ip
b
etwe
en
th
e
C
₂H₂
/C
₂H₄
an
d
C
₂H
6
/C
H₄
g
as r
atio
s
2
.
1
.
3
M
in
-
m
ax
no
rma
liza
t
io
n
Scale
f
ea
tu
r
es to
th
e
[
0
,
1
]
r
an
g
e
u
s
in
g
(
2
)
[
1
4
]
.
=
−
−
(
2
)
W
h
er
e
X
is
th
e
o
r
ig
in
al
v
alu
e,
an
d
ar
e
th
e
m
in
im
u
m
an
d
m
ax
im
u
m
v
alu
es
o
f
th
e
f
ea
tu
r
e,
r
esp
ec
tiv
ely
,
an
d
is
th
e
n
o
r
m
alize
d
v
alu
e.
T
h
is
tech
n
iq
u
e
s
tan
d
ar
d
izes
th
e
s
ca
les
o
f
g
as
r
atio
s
(
e.
g
.
,
C
H₄/H₂
,
C
₂H₂
/
C
₂H₄
)
,
en
s
u
r
in
g
th
eir
co
m
p
a
r
ab
i
lity
.
Fig
u
r
e
3
p
r
esen
ts
th
e
d
ataset
d
is
tr
ib
u
t
io
n
af
ter
ap
p
ly
in
g
m
in
–
m
ax
n
o
r
m
aliza
tio
n
.
Min
–
m
ax
n
o
r
m
aliza
tio
n
r
escales
g
as
r
atio
s
(
e.
g
.
,
C
H
₄
/
H
₂
,
C
₂
H
₆
/
CH
₄
)
to
th
e
[
0
,
1
]
r
an
g
e,
e
n
h
an
cin
g
th
eir
c
o
m
p
a
r
ab
ilit
y
,
as
illu
s
tr
ated
in
Fig
u
r
e
3
(
a)
.
H
o
wev
er
,
it
d
o
es
n
o
t
s
ig
n
if
ican
tly
r
e
d
u
ce
d
is
p
er
s
io
n
co
m
p
ar
e
d
to
r
aw
d
ata,
as
o
b
s
er
v
ed
in
Fig
u
r
e
3
(
b
)
,
wh
er
e
th
e
s
ca
tter
p
l
o
t
(
C
₂
H
₂
/
C
₂
H
₄
v
s
.
C
₂
H
₆
/
CH
₄
)
s
till
s
h
o
ws co
n
s
id
er
ab
le
class
o
v
er
lap
b
etwe
en
e
lectr
ical
f
au
lts
(
E
)
,
th
er
m
al
f
a
u
lts
(
T
)
,
an
d
n
o
r
m
a
l
co
n
d
itio
n
s
(
N)
.
Fig
u
r
e
3
(
a)
p
r
esen
ts
th
e
h
is
to
g
r
am
,
wh
ic
h
r
ev
ea
ls
m
o
r
e
d
is
tin
g
u
is
h
ab
le
a
v
er
ag
e
r
atio
s
,
with
C
₂
H
₆
/
CH
₄
d
o
m
in
an
t
f
o
r
elec
tr
ical
f
au
lts
(
E
)
an
d
C
H
₄
/
H
₂
f
o
r
th
er
m
al
f
au
lts
(
T
)
; n
ev
er
t
h
eless
,
clas
s
s
ep
ar
atio
n
r
em
ain
s
lim
ited
co
m
p
a
r
ed
to
t
h
e
lo
g
ar
ith
m
ic
t
r
an
s
f
o
r
m
ati
o
n
(
Fig
u
r
e
2
)
.
2
.
1
.
4
.
Sq
ua
re
ro
o
t
t
r
a
ns
f
o
rma
t
io
n
T
h
e
s
q
u
ar
e
r
o
o
t
tr
an
s
f
o
r
m
ati
o
n
h
ig
h
lig
h
ts
s
m
all
v
ar
iatio
n
s
wh
ile
atten
u
atin
g
lar
g
e
a
m
p
litu
d
es,
ac
co
r
d
in
g
to
(
3
)
.
′
=
√
√
(
3
)
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8
6
9
4
I
n
t J Po
w
E
lec
&
Dr
i Sy
s
t
,
Vo
l.
1
7
,
No
.
3
,
Sep
tem
b
er
20
2
6
:
178
0
-
1
7
9
3
1784
Fig
u
r
e
4
p
r
esen
ts
th
e
d
ataset
d
is
tr
ib
u
tio
n
af
ter
ap
p
ly
in
g
t
h
e
s
q
u
ar
e
r
o
o
t
tr
an
s
f
o
r
m
atio
n
.
T
h
e
s
q
u
ar
e
r
o
o
t
tr
an
s
f
o
r
m
atio
n
r
ed
u
ce
s
d
is
p
er
s
io
n
in
g
as
r
atio
s
(
e.
g
.
,
C
H
₄
/
H
₂
,
C
₂
H
₂
/
C
₂
H
₄
,
C
₂
H
₆
/
CH
₄
,
C
₂
H
₄
/
C
₂
H
₆
)
,
en
h
an
cin
g
co
m
p
ar
ab
ilit
y
ac
r
o
s
s
f
au
lt
ty
p
es,
as
s
h
o
wn
in
F
ig
u
r
e
4
.
I
n
t
h
e
b
a
r
c
h
ar
t
(
Fig
u
r
e
4
(
a
)
)
,
av
er
a
g
e
r
atio
s
ar
e
m
o
r
e
d
is
tin
g
u
is
h
ab
l
e,
with
C
₂
H
₆
/
CH
₄
p
r
o
m
in
en
t
f
o
r
elec
tr
ical
f
au
lts
(
E
)
an
d
C
₂
H
₂
/
C
₂
H
₄
elev
ated
f
o
r
t
h
er
m
al
f
au
lts
(
T
)
,
w
h
ile
n
o
r
m
al
s
tates
(
N)
s
h
o
w
b
alan
ce
d
r
atio
s
.
T
h
e
s
ca
tter
p
lo
t
(
Fig
u
r
e
4
(
b
)
)
(C
₂
H
₂
/
C
₂
H
₄
v
s
.
C
₂
H
₆
/
CH
₄
)
r
ev
ea
ls
im
p
r
o
v
ed
s
ep
a
r
atio
n
o
f
class
es
E
,
T
,
an
d
N
d
u
e
to
r
ed
u
ce
d
s
k
ewn
ess
,
s
im
ilar
to
lo
g
ar
ith
m
ic
tr
an
s
f
o
r
m
atio
n
,
th
er
e
b
y
en
h
an
cin
g
cla
s
s
if
ier
p
er
f
o
r
m
an
ce
.
(
a)
(
b
)
Fig
u
r
e
3.
Data
s
et
d
is
tr
ib
u
tio
n
(
m
in
-
m
ax
)
: (
a)
a
v
er
ag
e
g
as r
at
io
v
alu
es f
o
r
ea
ch
tr
a
n
s
f
o
r
m
er
f
au
lt c
ateg
o
r
y
,
an
d
(
b
)
s
ca
tter
p
lo
t illu
s
tr
atin
g
th
e
r
elatio
n
s
h
ip
b
etwe
en
th
e
C
₂H₂
/C
₂H₄
an
d
C
₂H
6
/C
H₄
g
as
r
atio
s
(
a)
(
b
)
Fig
u
r
e
4
.
Data
s
et
d
is
tr
ib
u
tio
n
(
Sq
u
ar
e
R
o
o
t)
:
(
a
)
av
er
a
g
e
g
as
r
atio
v
alu
es f
o
r
ea
ch
tr
a
n
s
f
o
r
m
er
f
au
lt c
ateg
o
r
y
,
an
d
(
b
)
s
ca
tter
p
lo
t illu
s
tr
atin
g
th
e
r
elatio
n
s
h
ip
b
etwe
en
th
e
C
₂H₂
/C
₂
H₄
an
d
C
₂H
6
/C
H₄
g
as
r
atio
s
2
.
2
.
M
a
chine
lea
rning
m
o
de
ls
T
h
is
s
tu
d
y
em
p
lo
y
s
f
o
u
r
s
u
p
er
v
is
ed
m
ac
h
in
e
lear
n
in
g
m
o
d
el
s
to
class
if
y
tr
an
s
f
o
r
m
er
f
au
lts
b
ased
o
n
DGA
g
as
r
atio
s
(
e.
g
.
,
C
H₄/H₂
,
C
₂H₂
/C
₂H₄
,
C
₂H₆
/
C
H₄,
C
₂H₄
/
C
₂H₆
)
:
R
F
[
1
5
]
,
[
1
6
]
,
SVM
[
1
7
]
,
g
r
ad
ien
t
-
b
o
o
s
te
d
tr
ee
s
(
GB
T
)
[
1
8
]
,
an
d
a
h
y
b
r
i
d
R
F
-
GB
T
m
o
d
el.
T
h
ese
m
o
d
els
wer
e
s
elec
ted
f
o
r
th
eir
r
o
b
u
s
tn
ess
,
ab
ilit
y
to
h
an
d
le
co
m
p
lex
d
ec
is
io
n
b
o
u
n
d
ar
ies,
an
d
ef
f
ec
tiv
e
n
ess
with
n
o
is
y
o
r
p
r
ep
r
o
ce
s
s
ed
d
ata
(
r
aw,
m
in
-
m
ax
,
lo
g
ar
ith
m
ic,
a
n
d
s
q
u
ar
e
r
o
o
t
tr
an
s
f
o
r
m
atio
n
s
)
.
SVM
with
an
R
B
F
k
er
n
el
is
well
-
s
u
ited
f
o
r
ca
p
tu
r
in
g
co
m
p
lex
n
o
n
lin
ea
r
d
ec
is
io
n
b
o
u
n
d
ar
ies
in
r
elativ
ely
s
m
all
d
atasets
.
R
F
is
k
n
o
wn
f
o
r
its
r
o
b
u
s
tn
e
s
s
to
n
o
is
e
an
d
its
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J Po
w
E
lec
&
Dr
i Sy
s
t
I
SS
N:
2088
-
8
6
9
4
Tr
a
n
s
fo
r
mer fa
u
lt d
ia
g
n
o
s
is
u
s
in
g
d
is
s
o
lved
g
a
s
a
n
a
lysi
s
:
a
h
yb
r
id
en
s
emb
le
…
(
F
a
tima
Z
o
h
r
a
B
o
u
d
jella
)
1785
ab
ilit
y
to
r
ed
u
ce
o
v
er
f
itti
n
g
t
h
r
o
u
g
h
en
s
em
b
le
a
v
er
ag
in
g
.
GB
T
,
o
n
th
e
o
th
er
h
a
n
d
,
p
r
o
v
id
es
h
ig
h
p
r
e
d
ictiv
e
ac
cu
r
ac
y
b
y
iter
ativ
ely
co
r
r
ec
t
in
g
p
r
ev
i
o
u
s
er
r
o
r
s
,
m
a
k
in
g
it
ef
f
ec
tiv
e
f
o
r
ca
p
tu
r
in
g
s
u
b
tle
p
atter
n
s
in
th
e
d
ata.
T
h
e
h
y
b
r
id
R
F
–
GB
T
m
o
d
el
is
p
r
o
p
o
s
ed
to
lev
er
a
g
e
t
h
e
s
ta
b
ilit
y
o
f
R
F
an
d
th
e
h
ig
h
ac
c
u
r
ac
y
o
f
GB
T
.
T
h
e
h
y
p
er
p
ar
am
eter
s
o
f
ea
ch
m
o
d
el
wer
e
s
elec
ted
b
ased
o
n
a
co
m
b
in
atio
n
o
f
e
m
p
ir
ical
tu
n
in
g
an
d
co
m
m
o
n
ly
ad
o
p
ted
co
n
f
ig
u
r
atio
n
s
r
e
p
o
r
t
ed
in
th
e
liter
atu
r
e.
Fo
r
th
e
S
VM
m
o
d
el,
th
e
R
B
F
k
er
n
el
w
as
ch
o
s
en
d
u
e
to
its
ab
ilit
y
to
m
o
d
el
n
o
n
lin
ea
r
r
el
atio
n
s
h
ip
s
,
wh
ile
th
e
r
eg
u
la
r
izatio
n
p
ar
a
m
eter
C
=
1
e
n
s
u
r
es
a
b
alan
ce
b
etwe
e
n
m
ar
g
in
m
a
x
im
izatio
n
a
n
d
clas
s
if
icatio
n
er
r
o
r
.
T
h
e
k
er
n
el
wi
d
th
p
ar
a
m
eter
σ
=
2
.
2
was
s
elec
ted
em
p
ir
ically
to
co
n
tr
o
l
th
e
i
n
f
lu
en
ce
o
f
in
d
iv
id
u
al
tr
ai
n
in
g
s
am
p
les.
Fo
r
th
e
R
F
m
o
d
el,
1
0
0
tr
ee
s
w
er
e
u
s
ed
to
en
s
u
r
e
s
u
f
f
icien
t
d
iv
er
s
ity
wh
ile
m
ai
n
tain
in
g
co
m
p
u
tatio
n
al
ef
f
icien
cy
.
T
h
e
m
ax
im
u
m
t
r
ee
d
ep
t
h
was
lim
ited
to
1
0
to
p
r
ev
en
t
o
v
er
f
itti
n
g
,
an
d
a
m
in
im
u
m
n
o
d
e
s
ize
o
f
2
was
s
elec
ted
to
allo
w
ad
eq
u
ate
m
o
d
el
f
lex
ib
ilit
y
.
T
h
e
Gin
i
in
d
ex
was
u
s
ed
as
th
e
s
p
l
it
cr
iter
io
n
d
u
e
to
its
ef
f
ec
tiv
e
n
ess
in
class
if
icatio
n
task
s
.
Fo
r
th
e
GB
T
m
o
d
el,
a
lar
g
e
n
u
m
b
e
r
o
f
tr
ee
s
(
1
0
,
0
0
0
)
co
m
b
in
ed
with
a
lear
n
in
g
r
a
te
o
f
0
.
1
was
u
s
ed
to
en
s
u
r
e
g
r
ad
u
al
lear
n
in
g
a
n
d
im
p
r
o
v
e
d
g
en
er
aliza
tio
n
.
T
h
e
tr
ee
d
ep
th
was r
estricte
d
to
2
to
r
ed
u
ce
m
o
d
el
co
m
p
lex
ity
an
d
av
o
id
o
v
er
f
itti
n
g
,
wh
ich
is
a
co
m
m
o
n
s
tr
ateg
y
i
n
b
o
o
s
tin
g
alg
o
r
ith
m
s
.
E
ac
h
m
o
d
el
was
tr
ain
ed
o
n
8
0
%
o
f
th
e
d
ataset
an
d
test
ed
o
n
2
0
%.
T
h
e
m
ac
h
in
e
lear
n
in
g
m
o
d
els
wer
e
im
p
lem
en
ted
u
s
in
g
th
e
KNI
ME
An
aly
tics
Platfo
r
m
with
th
e
h
y
p
er
p
ar
am
eter
s
s
u
m
m
ar
ize
d
in
T
ab
le
2
.
T
ab
le
2
.
Hy
p
er
p
ar
a
m
eter
s
ettin
g
s
f
o
r
t
h
e
class
if
icatio
n
m
o
d
e
ls
M
o
d
e
l
H
y
p
e
r
p
a
r
a
me
t
e
r
s
S
V
M
K
e
r
n
e
l
=
R
B
F
(
G
a
u
ssi
a
n
)
,
o
v
e
r
l
a
p
p
i
n
g
p
e
n
a
l
t
y
(
c
)
=
1
,
a
n
d
si
g
ma
(
σ)
=
2
.
2
RF
N
u
mb
e
r
o
f
t
r
e
e
s
=
1
0
0
,
sp
l
i
t
c
r
i
t
e
r
i
o
n
=
G
i
n
i
i
n
d
e
x
,
m
a
x
i
m
u
m
t
r
e
e
d
e
p
t
h
=
10
,
a
n
d
m
i
n
i
m
u
m
n
o
d
e
si
z
e
=
2
G
B
D
T
N
u
mb
e
r
o
f
t
r
e
e
s
=
1
0
0
0
0
,
l
e
a
r
n
i
n
g
r
a
t
e
=
0
.
1
,
a
n
d
ma
x
i
mu
m
t
r
e
e
d
e
p
t
h
=
2
2
.
2
.
1
.
RF
-
G
B
T
f
us
io
n
T
h
e
o
u
tp
u
ts
o
f
th
e
R
F
an
d
GB
T
class
if
ier
s
wer
e
co
m
b
in
ed
in
KNI
ME
u
s
in
g
th
e
"Pr
ed
icti
o
n
Fu
s
io
n
"
n
o
d
e.
T
h
is
f
u
s
io
n
a
p
p
r
o
ac
h
m
ak
es
u
s
e
o
f
R
F
'
s
r
o
b
u
s
tn
ess
an
d
in
ter
p
r
etab
ilit
y
as
w
ell
as
G
B
T
'
s
h
ig
h
p
r
ed
ictiv
e
ac
c
u
r
ac
y
an
d
g
lo
b
al
o
p
tim
izatio
n
ca
p
ab
ilit
ies.
A
weig
h
ted
p
r
o
b
a
b
ilit
y
av
e
r
a
g
in
g
tech
n
iq
u
e
was
u
s
ed
in
p
lace
o
f
eq
u
al
wei
g
h
tin
g
,
wh
ich
r
e
f
lecte
d
GB
T
'
s
b
etter
in
d
iv
id
u
al
p
e
r
f
o
r
m
an
ce
d
u
r
in
g
m
o
d
e
l
ev
alu
atio
n
.
Sp
ec
if
ically
,
th
e
f
u
s
ed
p
r
ed
ictio
n
was c
o
m
p
u
ted
as
(
4
)
.
(
,
)
=
2
3
(
,
)
+
1
3
(
,
)
(
4
)
T
h
e
f
in
al
class
lab
el
was d
eter
m
in
ed
b
y
(
5
)
.
̂
=
(
(
(
,
)
)
)
(
5
)
T
h
e
w
e
i
g
h
t
i
n
g
c
o
e
f
f
ic
i
e
n
ts
d
e
f
i
n
e
d
i
n
(
4
)
,
n
a
m
e
l
y
(
2
3
)
f
o
r
t
h
e
R
F
m
o
d
e
l
a
n
d
(
1
3
)
f
o
r
t
h
e
GB
T
m
o
d
e
l
,
w
e
r
e
d
e
t
e
r
m
i
n
e
d
t
h
r
o
u
g
h
a
n
e
m
p
i
r
i
c
a
l
g
r
i
d
-
s
e
a
r
c
h
p
r
o
c
e
d
u
r
e
d
u
r
i
n
g
t
h
e
v
a
l
i
d
a
t
i
o
n
p
h
a
s
e
b
y
s
y
s
t
e
m
a
t
i
ca
l
l
y
e
v
a
l
u
a
t
i
n
g
m
u
lt
i
p
l
e
w
e
i
g
h
ti
n
g
c
o
m
b
i
n
a
t
i
o
n
s
.
S
e
v
e
r
a
l
w
ei
g
h
t
in
g
c
o
m
b
i
n
a
t
i
o
n
s
w
e
r
e
t
es
t
e
d
i
n
o
r
d
e
r
t
o
as
s
e
s
s
t
h
ei
r
i
m
p
a
c
t
o
n
t
h
e
o
v
e
r
a
l
l
p
e
r
f
o
r
m
a
n
c
e
o
f
t
h
e
h
y
b
r
i
d
m
o
d
e
l
.
T
h
e
s
e
l
e
ct
e
d
c
o
n
f
i
g
u
r
a
t
i
o
n
p
r
o
v
i
d
e
d
t
h
e
b
e
s
t
t
r
a
d
e
-
o
f
f
b
e
t
w
e
e
n
a
c
c
u
r
a
c
y
a
n
d
r
o
b
u
s
t
n
e
s
s
.
T
h
e
h
i
g
h
e
r
w
e
i
g
h
t
as
s
i
g
n
ed
t
o
R
F
i
s
j
u
s
t
i
f
ie
d
b
y
i
t
s
a
b
i
l
ity
t
o
p
r
o
d
u
c
e
s
t
a
b
l
e
p
r
e
d
i
c
t
i
o
n
s
a
n
d
r
e
d
u
c
e
v
a
r
ia
n
ce
,
w
h
i
c
h
i
m
p
r
o
v
e
s
m
o
d
e
l
g
e
n
er
a
l
i
z
a
ti
o
n
.
I
n
c
o
n
t
r
as
t
,
G
B
T
,
a
lt
h
o
u
g
h
s
l
i
g
h
t
l
y
l
es
s
s
t
a
b
l
e
,
c
o
n
t
r
i
b
u
t
es
f
i
n
e
-
g
r
a
i
n
ed
d
i
s
c
r
i
m
i
n
at
i
v
e
p
o
w
e
r
t
h
r
o
u
g
h
i
t
s
s
e
q
u
e
n
t
i
al
l
e
a
r
n
i
n
g
p
r
o
c
e
s
s
.
W
it
h
t
h
e
g
o
a
l
o
f
l
o
w
e
r
i
n
g
c
l
as
s
i
f
i
c
at
i
o
n
e
r
r
o
r
s
a
n
d
b
o
o
s
t
i
n
g
t
h
e
r
o
b
u
s
t
n
es
s
o
f
t
h
e
d
i
a
g
n
o
s
t
i
c
s
y
s
t
e
m
,
t
h
i
s
w
ei
g
h
t
e
d
f
u
s
i
o
n
t
e
c
h
n
i
q
u
e
h
i
g
h
l
i
g
h
ts
t
h
e
m
o
r
e
a
c
c
u
r
a
t
e
GB
T
p
r
e
d
i
ct
i
o
n
s
w
h
i
le
p
r
e
s
e
r
v
i
n
g
t
h
e
e
n
s
e
m
b
l
e
a
d
v
an
t
a
g
e
s
o
f
R
F
.
2
.
3
.
E
x
perim
ent
a
l
a
pp
ro
a
ch
T
h
e
e
n
t
i
r
e
e
x
p
e
r
i
m
e
n
t
al
p
r
o
c
e
d
u
r
e
f
o
r
t
r
a
n
s
f
o
r
m
e
r
f
a
u
l
t d
i
a
g
n
o
s
i
s
u
s
i
n
g
D
GA
i
s
i
ll
u
s
t
r
at
e
d
in
F
i
g
u
r
e
5
.
Gas
d
ata
ar
e
f
ir
s
t
ex
tr
ac
ted
f
r
o
m
DGA
r
ep
o
r
ts
,
an
d
d
iag
n
o
s
tic
r
atio
s
,
p
ar
ticu
lar
ly
t
h
o
s
e
r
e
co
m
m
en
d
e
d
b
y
th
e
Mo
d
if
ied
R
o
g
er
s
’
R
atio
Me
th
o
d
,
a
r
e
co
m
p
u
ted
.
T
h
e
p
r
ep
r
o
ce
s
s
in
g
p
h
ase
in
clu
d
es
th
e
r
em
o
v
al
o
f
m
is
s
in
g
v
alu
es
an
d
th
e
a
p
p
licatio
n
o
f
f
o
u
r
d
ata
tr
a
n
s
f
o
r
m
atio
n
tech
n
iq
u
es:
r
aw
d
ata
(
u
s
ed
as
a
b
a
s
elin
e)
,
lo
g
ar
ith
m
ic
tr
an
s
f
o
r
m
atio
n
,
s
q
u
ar
e
r
o
o
t tr
a
n
s
f
o
r
m
atio
n
,
an
d
m
in
-
m
a
x
n
o
r
m
aliza
tio
n
.
T
h
e
KNI
ME
en
v
ir
o
n
m
e
n
t
is
th
en
u
s
ed
to
d
ev
elo
p
m
ac
h
i
n
e
lear
n
in
g
m
o
d
els,
s
u
ch
as
S
VM
,
R
F,
GB
T
,
an
d
a
h
y
b
r
id
R
F
-
GB
T
f
u
s
io
n
m
o
d
el.
E
ig
h
ty
p
er
ce
n
t
o
f
th
e
d
ataset
is
u
s
ed
to
tr
ain
e
ac
h
m
o
d
el,
a
n
d
t
h
e
r
em
ain
in
g
twen
ty
p
er
ce
n
t
is
u
s
ed
f
o
r
test
in
g
.
Fiv
e
co
m
m
o
n
m
etr
ics
ac
cu
r
ac
y
,
p
r
ec
is
io
n
,
r
ec
all,
F1
-
s
co
r
e,
an
d
C
o
h
en
'
s
k
ap
p
a
ar
e
u
s
ed
to
as
s
ess
th
e
m
o
d
el'
s
p
er
f
o
r
m
an
ce
.
T
h
is
p
r
o
c
ess
m
ak
es
it
p
o
s
s
ib
le
to
co
m
p
ar
e
t
h
e
m
o
d
els
in
g
r
ea
t d
etail
an
d
s
h
o
ws
h
o
w
v
ar
io
u
s
p
r
e
p
r
o
ce
s
s
in
g
tech
n
iq
u
es
af
f
ec
t
th
e
ac
cu
r
ac
y
an
d
d
ep
en
d
ab
ilit
y
o
f
th
e
f
i
n
al
d
iag
n
o
s
is
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8
6
9
4
I
n
t J Po
w
E
lec
&
Dr
i Sy
s
t
,
Vo
l.
1
7
,
No
.
3
,
Sep
tem
b
er
20
2
6
:
178
0
-
1
7
9
3
1786
Fig
u
r
e
5
.
E
x
p
er
im
e
n
tal
wo
r
k
f
l
o
w
f
o
r
tr
a
n
s
f
o
r
m
e
r
f
au
lt d
ia
g
n
o
s
is
3.
RE
SU
L
T
S AN
D
D
I
SCU
SS
I
O
N
T
h
e
r
esu
lts
o
f
th
e
class
if
icatio
n
m
o
d
els
ap
p
lied
to
DGA
-
b
ased
g
as
r
atio
d
ata
u
n
d
e
r
th
e
f
o
u
r
p
r
ep
r
o
ce
s
s
in
g
s
tr
ateg
ies
o
f
r
aw,
lo
g
ar
ith
m
ic,
s
q
u
ar
e
r
o
o
t,
an
d
m
i
n
-
m
ax
n
o
r
m
aliza
tio
n
ar
e
s
h
o
wn
in
th
is
s
ec
tio
n
.
SVM
with
R
B
F
k
er
n
el,
R
F,
G
B
T
,
an
d
a
h
y
b
r
id
R
F
-
GB
T
f
u
s
io
n
m
o
d
el
ar
e
am
o
n
g
th
e
m
o
d
els
th
at
h
av
e
b
ee
n
e
v
alu
ated
.
E
ig
h
ty
p
er
ce
n
t
o
f
th
e
d
ataset
was
u
s
ed
to
tr
ai
n
ea
ch
m
o
d
el,
a
n
d
t
h
e
r
em
ain
i
n
g
twen
t
y
p
er
ce
n
t
was
u
s
ed
f
o
r
test
in
g
.
Mo
d
el
p
er
f
o
r
m
an
ce
was
ev
al
u
ated
u
s
in
g
f
iv
e
k
ey
m
etr
ics:
ac
cu
r
ac
y
,
p
r
ec
is
io
n
,
r
ec
all,
F1
-
s
co
r
e,
an
d
C
o
h
en
’
s
k
ap
p
a
c
o
ef
f
icien
t.
T
h
e
r
esu
lts
ar
e
s
u
m
m
ar
ized
in
T
ab
le
3
.
T
h
e
R
F
class
if
ier
d
eliv
er
ed
s
tr
o
n
g
an
d
co
n
s
is
ten
t
p
e
r
f
o
r
m
an
ce
ac
r
o
s
s
all
p
r
ep
r
o
ce
s
s
in
g
s
tr
ateg
ies.
W
ith
r
aw
d
ata,
it
ac
h
ie
v
ed
7
9
.
5
%
ac
cu
r
ac
y
a
n
d
a
7
9
.
6
3
%
F1
-
s
co
r
e.
T
h
e
l
o
g
ar
ith
m
ic
tr
an
s
f
o
r
m
atio
n
led
to
t
h
e
b
est
r
esu
lts
f
o
r
R
F,
in
cr
ea
s
in
g
ac
cu
r
ac
y
to
8
3
.
3
%
an
d
C
o
h
en
’
s
k
ap
p
a
t
o
7
4
.
5
%.
B
o
th
s
q
u
ar
e
r
o
o
t
an
d
m
in
-
m
ax
tr
an
s
f
o
r
m
atio
n
s
also
p
r
o
v
id
ed
p
e
r
f
o
r
m
an
ce
g
ai
n
s
,
in
d
icatin
g
th
at
R
F
b
en
ef
its
f
r
o
m
r
ed
u
ce
d
s
k
ewn
ess
an
d
f
ea
t
u
r
e
s
ca
lin
g
.
T
h
is
b
eh
a
v
io
r
ca
n
b
e
ex
p
lain
ed
b
y
th
e
en
s
em
b
le
n
atu
r
e
o
f
R
F
b
ased
o
n
b
ag
g
in
g
,
w
h
ich
r
ed
u
ce
s
v
ar
ian
ce
a
n
d
m
ak
es
t
h
e
m
o
d
el
less
s
en
s
itiv
e
to
n
o
is
e
an
d
d
ata
d
is
tr
ib
u
tio
n
.
B
y
r
ed
u
cin
g
s
k
ewn
ess
,
lo
g
ar
ith
m
ic
a
n
d
s
q
u
ar
e
r
o
o
t
tr
an
s
f
o
r
m
atio
n
s
allo
w
m
o
r
e
b
alan
ce
d
tr
ee
s
p
lits
,
im
p
r
o
v
in
g
class
if
icatio
n
b
o
u
n
d
ar
ies.
GB
T
s
h
o
wed
s
tab
le
b
u
t
s
lig
h
tly
lo
wer
p
er
f
o
r
m
a
n
ce
th
an
R
F
ac
r
o
s
s
all
tr
an
s
f
o
r
m
atio
n
s
.
Acc
u
r
ac
y
r
em
ain
ed
ar
o
u
n
d
7
7
.
7
%
f
o
r
b
o
th
r
aw
an
d
m
in
-
m
a
x
s
ca
led
d
ata.
L
o
g
ar
ith
m
ic
an
d
s
q
u
ar
e
r
o
o
t
tr
an
s
f
o
r
m
atio
n
s
d
id
n
o
t
im
p
r
o
v
e
r
esu
lts
an
d
ev
en
ca
u
s
ed
s
lig
h
t
d
r
o
p
s
in
ac
cu
r
ac
y
an
d
k
ap
p
a.
T
h
is
s
u
g
g
ests
th
at
b
o
o
s
tin
g
alg
o
r
ith
m
s
m
ay
n
o
t
b
en
e
f
it
f
r
o
m
ex
ter
n
al
f
ea
tu
r
e
tr
an
s
f
o
r
m
atio
n
s
.
T
h
is
ca
n
b
e
ex
p
lain
ed
b
y
t
h
e
s
eq
u
en
tial
n
atu
r
e
o
f
GB
T
,
wh
ich
al
r
ea
d
y
p
er
f
o
r
m
s
in
ter
n
al
er
r
o
r
c
o
r
r
ec
tio
n
.
As
a
r
esu
lt,
e
x
ter
n
al
tr
an
s
f
o
r
m
atio
n
s
m
a
y
d
is
to
r
t
p
atter
n
s
th
at
th
e
m
o
d
el
is
alr
ea
d
y
ca
p
ab
le
o
f
ca
p
tu
r
i
n
g
,
esp
ec
ially
in
h
ig
h
ly
n
o
n
-
lin
ea
r
an
d
n
o
is
y
DGA
d
ata.
SVM
was
th
e
m
o
s
t
s
en
s
itiv
e
to
p
r
ep
r
o
ce
s
s
in
g
.
I
t
p
e
r
f
o
r
m
ed
m
o
d
er
ately
o
n
r
aw
d
at
a
(
6
8
.
5
%
ac
c
u
r
ac
y
)
,
b
u
t
d
r
am
atica
lly
u
n
d
e
r
p
er
f
o
r
m
ed
with
m
in
-
m
a
x
n
o
r
m
aliza
t
io
n
,
r
ea
ch
i
n
g
o
n
ly
3
3
.
8
%
ac
cu
r
ac
y
an
d
a
k
ap
p
a
o
f
1
%.
Ho
wev
er
,
with
lo
g
ar
ith
m
ic
tr
an
s
f
o
r
m
atio
n
,
SVM
ac
h
iev
ed
7
0
.
8
%
ac
cu
r
ac
y
an
d
a
m
o
r
e
b
alan
ce
d
F1
-
s
co
r
e
o
f
6
9
.
8
%.
T
h
ese
r
esu
lts
h
ig
h
lig
h
t
th
e
s
tr
o
n
g
s
en
s
itiv
ity
o
f
SVM
to
p
r
ep
r
o
ce
s
s
in
g
t
ec
h
n
iq
u
es.
T
h
is
is
m
ain
ly
d
u
e
to
its
r
elian
ce
o
n
d
is
tan
ce
-
b
ased
k
er
n
el
f
u
n
ctio
n
s
(
R
B
F),
wh
er
e
f
ea
tu
r
e
s
ca
li
n
g
d
ir
ec
tly
im
p
ac
ts
th
e
g
eo
m
etr
y
o
f
th
e
d
ec
is
io
n
s
p
ac
e.
L
o
g
ar
ith
m
ic
tr
an
s
f
o
r
m
atio
n
im
p
r
o
v
es
p
er
f
o
r
m
a
n
ce
b
y
s
tab
ilizin
g
v
ar
ian
ce
an
d
r
ed
u
cin
g
e
x
tr
em
e
v
alu
es,
m
ak
in
g
th
e
d
ata
m
o
r
e
s
u
itab
le
f
o
r
k
e
r
n
el
-
b
ased
s
ep
ar
atio
n
.
T
h
e
h
y
b
r
id
en
s
em
b
le
co
m
b
i
n
in
g
R
F
an
d
GB
T
o
u
tp
e
r
f
o
r
m
e
d
all
in
d
i
v
id
u
al
m
o
d
els.
O
n
r
aw
d
ata,
it
alr
ea
d
y
ac
h
iev
ed
9
0
.
3
%
ac
cu
r
ac
y
,
with
a
C
o
h
en
’
s
k
ap
p
a
o
f
8
1
.
5
%.
T
h
e
b
est
r
esu
lts
wer
e
o
b
tain
ed
with
th
e
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J Po
w
E
lec
&
Dr
i Sy
s
t
I
SS
N:
2088
-
8
6
9
4
Tr
a
n
s
fo
r
mer fa
u
lt d
ia
g
n
o
s
is
u
s
in
g
d
is
s
o
lved
g
a
s
a
n
a
lysi
s
:
a
h
yb
r
id
en
s
emb
le
…
(
F
a
tima
Z
o
h
r
a
B
o
u
d
jella
)
1787
lo
g
ar
ith
m
ic
tr
a
n
s
f
o
r
m
atio
n
,
p
ea
k
in
g
at
9
4
.
9
3
%
ac
cu
r
ac
y
an
d
9
4
.
8
1
%
F1
-
s
co
r
e,
alo
n
g
with
an
e
x
ce
llen
t
C
o
h
en
’
s
k
a
p
p
a
o
f
9
2
.
3
7
%,
in
d
icatin
g
n
ea
r
-
p
e
r
f
ec
t
a
g
r
ee
m
e
n
t
with
tr
u
e
la
b
els.
T
h
e
f
u
s
io
n
b
e
n
ef
its
f
r
o
m
th
e
co
m
p
lem
en
tar
y
s
tr
en
g
th
s
o
f
R
F (
lo
w
v
ar
ian
ce
)
a
n
d
GB
T
(
b
i
as c
o
r
r
ec
tio
n
)
,
r
esu
ltin
g
in
s
u
p
er
io
r
g
en
e
r
aliza
tio
n
an
d
r
o
b
u
s
tn
ess
ac
r
o
s
s
all
p
r
ep
r
o
ce
s
s
in
g
tech
n
iq
u
es.
T
h
is
co
n
f
ir
m
s
th
at
co
m
b
in
in
g
m
o
d
els
with
d
if
f
er
en
t
lear
n
in
g
m
ec
h
an
is
m
s
is
p
ar
ticu
lar
ly
ef
f
ec
tiv
e
f
o
r
h
an
d
li
n
g
th
e
co
m
p
lex
ity
a
n
d
v
ar
iab
ilit
y
o
f
DGA
d
ata.
T
h
is
p
er
f
o
r
m
an
ce
is
f
u
r
th
er
ill
u
s
tr
ated
b
y
th
e
co
n
f
u
s
io
n
m
at
r
ix
p
r
esen
ted
i
n
T
ab
le
4
,
wh
ich
p
r
o
v
i
d
es
a
d
etailed
v
iew
o
f
th
e
class
i
f
icatio
n
r
esu
lts
f
o
r
ea
ch
f
au
l
t
ca
teg
o
r
y
.
Ou
t
o
f
1
3
8
test
s
am
p
les,
1
3
1
wer
e
co
r
r
ec
tly
class
if
ied
,
co
r
r
esp
o
n
d
in
g
to
a
v
er
y
lo
w
er
r
o
r
r
ate
o
f
ap
p
r
o
x
im
ately
5
%.
As
s
h
o
wn
in
T
ab
le
4
,
th
e
m
o
d
el
ac
h
ie
v
ed
e
x
ce
llen
t
class
if
icatio
n
ac
cu
r
ac
y
f
o
r
th
e
n
o
r
m
al
(
N)
an
d
e
lectr
ical
(
E
)
class
es,
with
4
8
o
u
t
o
f
4
9
n
o
r
m
al
s
am
p
les an
d
4
5
o
u
t
o
f
4
8
elec
tr
ical
f
au
lt sam
p
les c
o
r
r
ec
tly
id
e
n
tifie
d
.
T
h
e
t
h
e
r
m
al
(
T
)
cl
ass
als
o
s
h
o
ws
s
t
r
o
n
g
p
e
r
f
o
r
m
a
n
c
e,
wit
h
3
8
c
o
r
r
ec
t
ly
class
if
ie
d
s
a
m
p
les
o
u
t
o
f
4
1
.
Ho
w
ev
er
,
m
o
s
t
m
is
cl
ass
i
f
ic
at
io
n
s
o
cc
u
r
b
etw
ee
n
t
h
er
m
a
l
f
a
u
lts
a
n
d
n
o
r
m
al
s
ta
tes
.
S
p
ec
i
f
ic
all
y
,
T
ab
le
4
in
d
ic
at
es
t
h
at
2
t
h
e
r
m
al
f
a
u
lt
c
ases
we
r
e
i
n
c
o
r
r
ec
t
ly
p
r
e
d
ic
te
d
as
n
o
r
m
al,
an
d
1
n
o
r
m
al
ca
s
e
w
as
m
is
class
if
i
e
d
as
t
h
e
r
m
al
.
T
h
is
c
o
n
f
u
s
i
o
n
ca
n
b
e
att
r
i
b
u
te
d
t
o
m
o
d
e
r
at
e
t
h
e
r
m
al
d
e
g
r
a
d
a
ti
o
n
c
o
n
d
iti
o
n
s
,
w
h
e
r
e
g
as
r
ati
o
s
r
e
m
ai
n
cl
o
s
e
t
o
t
h
o
s
e
o
b
s
e
r
v
e
d
i
n
h
e
alt
h
y
tr
an
s
f
o
r
m
er
s
.
Ad
d
i
t
io
n
al
ly
,
a
s
m
all
n
u
m
b
e
r
o
f
m
is
class
i
f
i
ca
t
io
n
s
c
an
b
e
o
b
s
er
v
e
d
b
etw
ee
n
ele
ct
r
ic
al
a
n
d
t
h
e
r
m
al
f
au
lts
(
T
a
b
le
4
)
,
s
u
g
g
esti
n
g
p
a
r
ti
al
o
v
er
l
a
p
i
n
ce
r
ta
in
b
o
r
d
e
r
li
n
e
g
as
p
at
te
r
n
s
.
T
h
is
p
h
en
o
m
e
n
o
n
r
e
f
lects
t
h
e
p
h
y
s
i
ca
l
r
e
alit
y
o
f
tr
a
n
s
f
o
r
m
er
b
e
h
a
v
i
o
r
,
w
h
e
r
e
f
a
u
l
t
s
i
g
n
at
u
r
es
a
r
e
n
o
t
alwa
y
s
s
h
ar
p
l
y
s
e
p
a
r
a
b
l
e,
es
p
ec
i
all
y
in
e
ar
ly
f
a
u
lt
s
ta
g
es
.
F
r
o
m
a
n
ap
p
l
ica
ti
o
n
p
e
r
s
p
ec
ti
v
e,
th
ese
r
es
u
lts
a
r
e
s
ig
n
i
f
ic
a
n
t.
T
h
e
l
o
w
n
u
m
b
e
r
o
f
m
is
c
lass
i
f
i
ca
t
io
n
s
a
n
d
t
h
e
h
i
g
h
a
cc
u
r
ac
y
f
o
r
cr
itic
al
f
au
lt
c
lass
es
d
em
o
n
s
t
r
at
e
th
a
t
t
h
e
p
r
o
p
o
s
e
d
R
F
–
GB
T
m
o
d
el
is
r
e
lia
b
l
e
f
o
r
r
ea
l
-
w
o
r
l
d
tr
a
n
s
f
o
r
m
e
r
m
o
n
it
o
r
i
n
g
.
I
n
p
a
r
tic
u
la
r
,
m
i
n
i
m
i
zi
n
g
co
n
f
u
s
io
n
b
etw
ee
n
n
o
r
m
a
l
an
d
f
a
u
l
ty
c
o
n
d
iti
o
n
s
is
ess
en
t
ial
f
o
r
r
e
d
u
c
in
g
f
alse
al
ar
m
s
an
d
a
v
o
id
in
g
u
n
n
ec
ess
a
r
y
m
ai
n
t
e
n
a
n
c
e
ac
ti
o
n
s
.
O
v
e
r
a
ll
,
T
a
b
l
e
4
co
n
f
i
r
m
s
n
o
t
o
n
l
y
t
h
e
h
i
g
h
g
lo
b
a
l
ac
c
u
r
ac
y
o
f
t
h
e
m
o
d
el
,
b
u
t
als
o
its
s
tr
o
n
g
class
-
wis
e
p
er
f
o
r
m
a
n
c
e
a
n
d
p
r
ac
t
ic
al
s
u
i
ta
b
il
it
y
f
o
r
DGA
-
b
as
ed
f
a
u
lt
d
ia
g
n
o
s
is
.
T
h
is
m
eth
o
d
o
l
o
g
ical
ch
o
ice
was
m
ad
e
to
en
s
u
r
e
r
eliab
le
an
d
b
alan
ce
d
lear
n
in
g
d
u
r
i
n
g
m
o
d
el
tr
ain
in
g
.
T
h
e
co
llected
DGA
r
ep
o
r
ts
p
r
esen
te
d
h
eter
o
g
e
n
e
o
u
s
d
ata
av
ailab
ilit
y
,
an
d
d
et
ailed
s
u
b
ca
teg
o
r
ies
s
u
ch
as
T
1
/T2
/T3
o
r
PD/D1
/D2
wer
e
n
o
t
s
y
s
tem
atica
lly
r
ep
o
r
te
d
in
in
d
u
s
tr
ial
r
ec
o
r
d
s
.
I
n
clu
d
i
n
g
th
ese
s
u
b
class
es
wo
u
ld
h
av
e
r
esu
lt
ed
in
s
ig
n
if
ican
t
class
im
b
ala
n
ce
an
d
r
e
d
u
ce
d
th
e
s
tatis
tica
l
r
o
b
u
s
tn
ess
o
f
th
e
m
o
d
els.
T
h
er
ef
o
r
e,
th
e
ad
o
p
ted
th
r
ee
-
class
s
tr
u
ctu
r
e
p
r
o
v
id
es
a
p
r
ac
tical
an
d
g
en
er
aliza
b
le
d
iag
n
o
s
tic
f
r
am
ewo
r
k
,
ca
p
ab
le
o
f
s
u
p
p
o
r
tin
g
c
o
n
d
itio
n
-
b
ased
m
ai
n
ten
an
ce
d
ec
is
io
n
s
wh
ile
m
ain
tain
in
g
s
tr
o
n
g
p
r
ed
ictiv
e
s
tab
ilit
y
an
d
i
n
ter
p
r
etab
ilit
y
.
T
ab
le
3
.
C
o
m
p
a
r
is
o
n
o
f
class
i
f
ier
p
er
f
o
r
m
an
ce
ac
co
r
d
in
g
to
d
ata
tr
an
s
f
o
r
m
atio
n
tech
n
i
q
u
e
s
C
l
a
s
si
f
i
e
r
D
a
t
a
A
c
c
u
r
a
c
y
(
%)
P
r
e
c
i
s
i
o
n
(
%)
R
e
c
a
l
l
(
%)
F
-
mea
su
r
e
(
%)
C
o
h
e
n
’
s
k
a
p
p
a
(
%)
RF
R
a
w
d
a
t
a
79
.
5
79
.
93
79
.
53
79
.
63
71
.
3
M
i
n
-
ma
x
80
.
5
81
.
07
80
.
53
80
.
7
71
.
3
LO
G
83
.
3
83
.
66
83
.
3
83
.
4
74
.
5
S
q
u
a
r
e
r
o
o
t
82
.
3
82
.
67
82
.
3
82
.
4
73
.
5
G
B
T
R
a
w
d
a
t
a
77
.
7
78
.
163
77
.
63
77
.
4
66
.
5
M
i
n
-
ma
x
77
.
7
78
.
17
77
.
63
77
.
4
66
.
5
LO
G
76
.
9
77
.
16
76
.
86
76
.
63
65
.
4
S
q
u
a
r
e
r
o
o
t
76
.
9
77
.
17
76
.
87
76
.
2
76
.
63
S
V
M
R
a
w
d
a
t
a
68
.
5
72
.
13
68
.
4
68
.
03
52
.
7
M
i
n
-
ma
x
33
.
8
25
.
25
34
.
07
29
.
85
1
LO
G
7
0
,
8
71
.
43
70
.
93
69
.
8
56
.
2
S
q
u
a
r
e
r
o
o
t
51
.
5
51
.
87
51
.
57
51
.
27
27
.
3
RF
-
G
B
T
R
a
w
d
a
t
a
90
.
3
90
.
66
90
.
3
90
.
33
81
.
5
M
i
n
-
ma
x
92
.
3
92
.
33
92
.
3
92
.
33
83
.
5
LO
G
94
.
93
94
.
88
94
.
8
94
.
81
92
.
37
S
q
u
a
r
e
r
o
o
t
93
.
8
94
.
03
93
.
93
93
.
93
90
.
8
T
ab
le
4
.
C
o
n
f
u
s
io
n
m
atr
ix
f
o
r
th
e
R
F
-
GB
T
m
o
d
el
with
lo
g
ar
ith
m
ic
tr
an
s
f
o
r
m
atio
n
N
u
mb
e
r
o
f
r
o
w
s
:
1
3
8
E
(
P
r
e
d
i
c
t
e
d
)
N
(
P
r
e
d
i
c
t
e
d
)
T
(
P
r
e
d
i
c
t
e
d
)
E
(
A
c
t
u
a
l
)
45
1
2
N
(
A
c
t
u
a
l
)
0
48
1
T
(
A
c
t
u
a
l
)
1
2
38
3
.
1
.
Co
m
pa
ra
t
iv
e
a
na
ly
s
is
o
f
t
he
m
o
dels
T
h
e
g
r
ap
h
(
F
ig
u
r
e
6
)
p
r
o
v
id
e
s
a
co
m
p
ar
ativ
e
ev
alu
atio
n
o
f
f
o
u
r
class
if
ier
s
:
R
F,
G
B
T
,
SVM,
an
d
th
eir
en
s
em
b
le
co
m
b
in
atio
n
(
R
F
-
GB
T
)
,
b
ased
o
n
f
iv
e
k
ey
p
er
f
o
r
m
an
ce
in
d
icato
r
s
.
T
h
e
c
lass
if
icatio
n
r
esu
lt
s
b
ased
o
n
th
e
f
in
al
ev
alu
atio
n
m
etr
ics
clea
r
ly
d
em
o
n
s
tr
ate
th
at
th
e
R
F
-
G
B
T
en
s
em
b
le
m
o
d
el
s
ig
n
if
ican
tly
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8
6
9
4
I
n
t J Po
w
E
lec
&
Dr
i Sy
s
t
,
Vo
l.
1
7
,
No
.
3
,
Sep
tem
b
er
20
2
6
:
178
0
-
1
7
9
3
1788
o
u
tp
er
f
o
r
m
s
all
o
th
er
class
if
ier
s
.
W
ith
an
ac
cu
r
ac
y
o
f
9
2
.
6
%,
F1
-
s
co
r
e
o
f
9
2
.
4
2
%,
an
d
a
C
o
h
en
’
s
k
a
p
p
a
o
f
8
6
.
8
%,
it
d
eliv
er
s
th
e
m
o
s
t
r
e
liab
le
an
d
c
o
n
s
is
ten
t
p
r
ed
ictio
n
s
.
T
h
is
h
y
b
r
id
a
p
p
r
o
ac
h
lev
e
r
ag
es
th
e
v
a
r
ian
ce
co
n
tr
o
l
o
f
R
F
an
d
th
e
b
ias
co
r
r
ec
tio
n
s
tr
en
g
t
h
o
f
GB
T
,
r
esu
ltin
g
in
s
u
p
er
io
r
g
en
er
aliza
tio
n
an
d
f
au
lt
d
is
cr
im
in
atio
n
.
T
h
e
s
tan
d
alo
n
e
R
F
m
o
d
el
f
o
llo
ws
with
8
1
.
4
%
ac
cu
r
ac
y
,
8
1
.
5
3
%
F1
-
s
co
r
e,
an
d
a
C
o
h
e
n
’
s
k
ap
p
a
o
f
7
2
.
7
%.
T
h
ese
r
esu
lts
r
ef
lect
a
r
o
b
u
s
t
a
n
d
s
tab
le
class
if
ier
th
at
h
an
d
les
n
o
is
y
a
n
d
v
ar
iab
le
d
ata
well,
m
ak
in
g
it a
s
o
lid
o
p
tio
n
w
h
en
m
o
d
el
s
im
p
licity
o
r
i
n
ter
p
r
eta
b
ilit
y
is
r
eq
u
ir
ed
.
T
h
e
GB
T
p
er
f
o
r
m
m
o
d
er
ately
,
with
7
7
.
3
%
ac
cu
r
ac
y
,
7
7
.
1
6
%
F1
-
s
co
r
e,
an
d
6
8
.
7
%
o
n
C
o
h
en
’
s
k
ap
p
a.
W
h
ile
ef
f
ec
tiv
e,
th
e
m
o
d
el
f
alls
s
h
o
r
t
o
f
R
F
in
b
o
th
p
r
ec
is
io
n
a
n
d
r
eliab
ilit
y
,
p
o
s
s
ib
ly
d
u
e
t
o
o
v
er
f
itti
n
g
o
r
s
en
s
itiv
ity
to
s
p
ec
if
ic
d
ata
p
atter
n
s
.
T
h
e
SVM
m
o
d
el
d
eliv
er
s
th
e
wea
k
est r
e
s
u
lts
,
with
ac
cu
r
ac
y
lim
ited
to
5
6
.
1
5
%,
F1
-
s
co
r
e
at
5
4
.
2
4
%,
an
d
a
lo
w
C
o
h
en
’
s
k
ap
p
a
o
f
3
4
.
3
%.
T
h
is
h
ig
h
lig
h
ts
its
p
o
o
r
alig
n
m
en
t
with
ac
tu
al
f
au
lt
class
es
an
d
s
u
g
g
ests
lim
ited
s
u
ita
b
ilit
y
f
o
r
DGA
-
b
ased
tr
an
s
f
o
r
m
er
d
iag
n
o
s
tics
in
its
cu
r
r
en
t c
o
n
f
ig
u
r
atio
n
.
Fig
u
r
e
6
.
C
o
m
p
a
r
is
o
n
o
f
p
r
ec
i
s
io
n
,
r
ec
all,
F
-
m
ea
s
u
r
e,
ac
c
u
r
a
cy
,
an
d
C
o
h
en
s
co
r
es f
o
r
ea
ch
class
if
ier
3
.
2
.
Co
m
pa
riso
n o
f
perf
o
rma
nce
ba
s
ed
o
n da
t
a
t
ra
ns
f
o
r
m
a
t
io
ns
(
n
o
r
m
a
liza
t
io
n)
T
o
b
etter
is
o
late
th
e
im
p
ac
t
o
f
p
r
ep
r
o
ce
s
s
in
g
m
eth
o
d
s
o
n
th
e
m
o
s
t
co
n
s
is
ten
t
class
if
ier
s
,
th
e
an
aly
s
is
f
o
cu
s
ed
s
o
lely
o
n
R
F,
GB
T
,
a
n
d
th
eir
h
y
b
r
id
en
s
em
b
le
(
R
F
-
GB
T
)
,
ex
clu
d
i
n
g
SVM
d
u
e
to
its
in
s
tab
ilit
y
(
s
ee
F
ig
u
r
e
7
)
.
B
ased
o
n
a
v
er
ag
e
p
er
f
o
r
m
an
ce
ac
r
o
s
s
th
ese
m
o
d
els,
lo
g
ar
ith
m
ic
tr
an
s
f
o
r
m
atio
n
y
ield
ed
t
h
e
b
est
o
v
er
all
r
esu
lts
,
with
an
av
er
a
g
e
ac
cu
r
ac
y
o
f
8
4
.
4
%,
F1
-
s
co
r
e
o
f
8
4
.
3
%,
an
d
C
o
h
e
n
’
s
k
ap
p
a
o
f
8
0
%.
I
t
clea
r
l
y
im
p
r
o
v
es th
e
lear
n
in
g
p
r
o
ce
s
s
b
y
r
ed
u
cin
g
s
k
ewn
ess
an
d
s
tab
ilizin
g
f
ea
tu
r
e
d
is
tr
ib
u
tio
n
s
.
Fig
u
r
e
7
.
C
o
m
p
a
r
is
o
n
o
f
av
er
a
g
e
p
er
f
o
r
m
an
ce
s
ac
c
o
r
d
in
g
to
d
ata
tr
an
s
f
o
r
m
atio
n
tech
n
i
q
u
e
s
Sq
u
ar
e
r
o
o
t
tr
an
s
f
o
r
m
atio
n
p
r
o
d
u
ce
d
alm
o
s
t
id
en
tical
r
esu
lts
(
ac
cu
r
ac
y
:
8
4
.
3
%,
F1
-
s
co
r
e:
8
4
.
1
8
%,
k
ap
p
a:
8
0
.
2
%),
co
n
f
ir
m
i
n
g
its
s
u
itab
ilit
y
f
o
r
th
is
class
if
icati
o
n
task
.
T
h
e
n
eg
lig
ib
le
d
if
f
er
en
ce
s
u
g
g
ests
b
o
th
tr
an
s
f
o
r
m
atio
n
s
ar
e
ef
f
ec
tiv
e
,
with
s
q
u
ar
e
r
o
o
t
o
f
f
e
r
in
g
s
lig
h
tly
b
etter
ag
r
ee
m
en
t
with
tr
u
e
lab
els.
Min
-
m
ax
n
o
r
m
aliza
tio
n
led
to
a
m
o
d
est
im
p
r
o
v
em
en
t
o
v
e
r
r
aw
d
ata,
with
ac
cu
r
ac
y
in
cr
ea
s
in
g
f
r
o
m
8
2
.
5
%
to
8
3
.
5
%,
an
d
F1
-
s
co
r
e
f
r
o
m
8
2
.
4
5
%
to
8
3
.
4
8
%.
Ho
wev
er
,
th
e
im
p
ac
t
o
n
C
o
h
en
’
s
k
ap
p
a
r
em
ain
s
li
m
ited
(
f
r
o
m
7
3
.
1
%
to
7
3
.
8
%),
i
n
d
icatin
g
m
in
o
r
in
f
lu
en
ce
o
n
lab
el
co
n
s
is
ten
cy
.
Ov
er
all,
th
e
f
in
d
in
g
s
s
u
p
p
o
r
t
th
e
u
s
e
o
f
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J Po
w
E
lec
&
Dr
i Sy
s
t
I
SS
N:
2088
-
8
6
9
4
Tr
a
n
s
fo
r
mer fa
u
lt d
ia
g
n
o
s
is
u
s
in
g
d
is
s
o
lved
g
a
s
a
n
a
lysi
s
:
a
h
yb
r
id
en
s
emb
le
…
(
F
a
tima
Z
o
h
r
a
B
o
u
d
jella
)
1789
lo
g
ar
ith
m
ic
o
r
s
q
u
ar
e
r
o
o
t
tr
a
n
s
f
o
r
m
atio
n
as
th
e
m
o
s
t
ef
f
ec
tiv
e
p
r
e
p
r
o
ce
s
s
in
g
s
tr
ateg
ies,
wh
ile
r
aw
a
n
d
m
in
-
m
ax
s
ca
led
d
ata
r
em
ai
n
ac
ce
p
t
ab
le
b
u
t le
s
s
o
p
tim
al.
3
.
3
.
Abla
t
io
n
s
t
ud
y
:
im
pa
c
t
o
f
DG
A
r
a
t
io
s
T
o
ev
alu
ate
th
e
c
o
n
tr
ib
u
tio
n
o
f
d
if
f
er
e
n
t
f
ea
t
u
r
es
d
er
i
v
e
d
f
r
o
m
DGA,
a
n
a
b
latio
n
s
tu
d
y
was
co
n
d
u
cte
d
.
T
h
e
o
b
jectiv
e
wa
s
to
an
aly
ze
h
o
w
v
ar
io
u
s
c
o
m
b
in
atio
n
s
o
f
g
as
r
ati
o
s
in
f
l
u
en
ce
th
e
d
iag
n
o
s
tic
p
er
f
o
r
m
an
ce
o
f
t
h
e
p
r
o
p
o
s
ed
m
o
d
el.
T
h
e
r
esu
lts
o
f
th
is
a
n
aly
s
is
ar
e
s
u
m
m
ar
ized
in
T
a
b
le
5
an
d
ar
e
d
is
cu
s
s
ed
in
d
etail
to
h
ig
h
lig
h
t
th
e
c
o
n
tr
ib
u
tio
n
o
f
ea
c
h
f
ea
tu
r
e
c
o
m
b
in
atio
n
.
Fo
r
th
is
ex
p
e
r
im
en
t
,
th
e
f
u
s
io
n
m
o
d
el
co
m
b
in
in
g
GB
T
an
d
R
F
was u
s
ed
,
as it
ac
h
iev
ed
th
e
b
est o
v
er
all
p
er
f
o
r
m
a
n
ce
in
th
e
p
r
ev
i
o
u
s
ex
p
er
im
en
ts
.
I
n
ad
d
itio
n
,
lo
g
ar
ith
m
ic
n
o
r
m
ali
za
tio
n
was
ap
p
lied
to
th
e
in
p
u
t
d
ata
s
in
ce
it
p
r
o
v
id
ed
th
e
m
o
s
t
ef
f
ec
tiv
e
r
esu
lts
co
m
p
ar
ed
with
o
th
er
n
o
r
m
aliza
tio
n
tech
n
iq
u
es.
Fo
u
r
f
ea
tu
r
e
co
n
f
ig
u
r
atio
n
s
b
ased
o
n
d
if
f
er
en
t
co
m
b
i
n
atio
n
s
o
f
DGA
r
atio
s
wer
e
in
v
esti
g
at
ed
to
ev
alu
ate
t
h
eir
im
p
ac
t
o
n
th
e
d
iag
n
o
s
tic
m
o
d
el
p
er
f
o
r
m
a
n
ce
.
T
h
e
r
esu
lts
s
h
o
w
th
at
u
s
in
g
o
n
ly
th
e
C
H₄/H₂
r
atio
p
r
o
v
id
es
a
m
o
d
er
ate
b
aselin
e
p
er
f
o
r
m
an
c
e
with
an
ac
cu
r
ac
y
o
f
8
5
.
4
%.
As
s
h
o
wn
in
T
ab
le
5
,
th
is
r
atio
ca
p
tu
r
es
ess
en
tia
l
in
f
o
r
m
atio
n
r
elate
d
to
th
er
m
al
f
au
lts
,
b
u
t
r
em
ain
s
in
s
u
f
f
icien
t
to
f
u
ll
y
d
is
cr
im
in
ate
b
etwe
en
d
if
f
er
e
n
t
f
au
lt
ty
p
es.
Ho
wev
er
,
th
e
c
o
m
b
in
atio
n
o
f
two
r
atio
s
s
lig
h
tly
d
ec
r
ea
s
es
th
e
p
er
f
o
r
m
a
n
ce
,
in
d
icatin
g
th
at
th
es
e
two
f
ea
tu
r
es
alo
n
e
ar
e
n
o
t
s
u
f
f
icien
t
to
ca
p
tu
r
e
th
e
co
m
p
lex
ity
o
f
tr
an
s
f
o
r
m
er
f
a
u
lt
p
atter
n
s
.
T
h
is
d
e
cr
ea
s
e
s
u
g
g
ests
p
o
s
s
ib
le
r
e
d
u
n
d
a
n
cy
o
r
wea
k
co
m
p
lem
en
tar
ity
b
etwe
en
th
e
s
elec
ted
r
atio
s
,
wh
ich
m
a
y
in
tr
o
d
u
ce
n
o
is
e
r
ath
e
r
th
a
n
im
p
r
o
v
in
g
class
s
ep
ar
ab
ilit
y
.
A
s
ig
n
if
ican
t
im
p
r
o
v
em
en
t
is
o
b
s
er
v
ed
wh
e
n
th
r
ee
r
atio
s
ar
e
u
s
ed
,
wh
er
e
th
e
ac
cu
r
ac
y
in
cr
ea
s
es
to
9
4
.
6
%
an
d
C
o
h
en
’
s
k
ap
p
a
r
ea
ch
es
9
1
.
9
%,
d
em
o
n
s
tr
atin
g
th
at
th
e
a
d
d
itio
n
al
r
atio
co
n
tr
ib
u
tes
m
ea
n
in
g
f
u
l
d
iag
n
o
s
tic
in
f
o
r
m
atio
n
.
Acc
o
r
d
in
g
to
T
ab
le
5
,
th
is
co
n
f
i
g
u
r
atio
n
p
r
o
v
id
es
a
b
etter
b
alan
c
e
b
etwe
en
f
ea
tu
r
e
d
iv
er
s
ity
an
d
d
is
cr
im
in
ativ
e
p
o
wer
,
en
ab
lin
g
th
e
m
o
d
el
to
c
ap
tu
r
e
m
o
r
e
co
m
p
lex
f
a
u
lt si
g
n
atu
r
es.
T
h
e
h
ig
h
est
p
er
f
o
r
m
an
ce
is
ac
h
ie
v
ed
with
f
o
u
r
r
atio
s
,
r
ea
c
h
in
g
an
ac
c
u
r
ac
y
o
f
9
4
.
9
3
%
a
n
d
a
C
o
h
en
’
s
k
ap
p
a
o
f
9
2
.
3
7
%.
T
h
is
r
esu
lt,
clea
r
ly
h
ig
h
lig
h
ted
in
T
a
b
le
5
,
co
n
f
ir
m
s
th
at
co
m
b
in
i
n
g
m
u
ltip
le
co
m
p
le
m
en
tar
y
g
as
r
atio
s
s
ig
n
if
ican
tly
en
h
an
ce
s
th
e
m
o
d
el’
s
ab
ilit
y
to
d
is
tin
g
u
is
h
b
e
twee
n
f
au
lt
ty
p
es.
Fro
m
a
p
r
ac
tical
p
er
s
p
ec
tiv
e,
th
ese
f
in
d
in
g
s
in
d
icate
th
at
r
ely
in
g
o
n
a
s
in
g
le
o
r
lim
ited
n
u
m
b
er
o
f
g
as
r
atio
s
m
ay
l
ea
d
to
in
co
m
p
lete
d
iag
n
o
s
is
,
wh
ile
in
co
r
p
o
r
atin
g
m
u
ltip
le
r
atio
s
im
p
r
o
v
es
r
o
b
u
s
tn
ess
an
d
r
eliab
ilit
y
in
r
ea
l
-
wo
r
ld
tr
an
s
f
o
r
m
e
r
m
o
n
ito
r
in
g
s
y
s
tem
s
.
T
h
is
co
n
f
ir
m
s
th
at
in
co
r
p
o
r
atin
g
m
u
lti
p
le
DGA
r
atio
s
en
h
an
ce
s
th
e
m
o
d
el’
s
ab
ilit
y
to
ca
p
tu
r
e
r
elev
a
n
t f
au
lt c
h
ar
ac
te
r
is
tics
an
d
im
p
r
o
v
es th
e
r
eliab
ilit
y
o
f
tr
an
s
f
o
r
m
er
f
a
u
lt d
iag
n
o
s
is
.
T
ab
le
5
.
Ab
latio
n
s
tu
d
y
r
esu
lts
f
o
r
d
if
f
e
r
en
t c
o
m
b
in
atio
n
s
o
f
DGA
r
atio
s
F
e
a
t
u
r
e
s
A
c
c
u
r
a
c
y
(
%)
P
r
e
c
i
s
i
o
n
(
%)
R
e
c
a
l
l
(
%)
F
-
mesu
r
e
(
%)
C
o
h
e
n
’
s
k
a
p
p
a
(
%)
C
H
4
/
H
2
85
.
4
86
.
13
86
.
26
85
.
4
78
.
2
C
H
4
/
H
2
,
C
2
H
2
/
C
2
H
4
83
.
1
82
.
5
83
.
63
82
.
66
74
.
5
C
H
4
/
H
2
,
C
2
H
2
/
C
2
H
4
,
C
2
H
4
/
C
2
H
6
94
.
6
94
.
2
95
.
16
94
.
4
91
.
9
C
H
4
/
H
2
,
C
2
H
2
/
C
2
H
4
,
C
2
H
4
/
C
2
H
6
,
C
2
H
2
/
C
H
4
94
.
93
94
.
88
94
.
8
94
.
81
92
.
37
3
.
4
.
Co
mp
a
ra
t
iv
e
ev
a
lua
t
io
n wit
h t
ra
ditio
na
l
DG
A
inte
r
pr
et
a
t
io
n met
ho
ds
T
o
ass
es
s
th
e
r
elev
an
ce
o
f
th
e
p
r
o
p
o
s
ed
h
y
b
r
id
m
o
d
el,
a
d
ir
ec
t
co
m
p
ar
is
o
n
was
ca
r
r
ied
o
u
t
with
th
e
m
o
s
t
co
m
m
o
n
ly
u
s
ed
tr
a
d
itio
n
al
DGA
in
ter
p
r
etatio
n
m
eth
o
d
s
:
th
e
Du
v
al
T
r
ian
g
le,
th
e
R
o
g
er
s
R
atio
Me
th
o
d
,
an
d
th
e
I
E
C
6
0
5
9
9
r
atio
m
eth
o
d
.
T
h
e
d
ef
ec
t
s
u
b
ca
teg
o
r
ies we
r
e
g
r
o
u
p
e
d
as f
o
llo
ws:
(
T
1
+
T
2
+
T
3
→
T
)
an
d
(
PD
+
D1
+
D2
→
E
)
.
T
h
e
co
m
p
ar
ativ
e
r
esu
lts
p
r
esen
ted
in
T
ab
le
6
s
h
o
w
th
at
co
n
v
en
tio
n
al
ap
p
r
o
ac
h
es
ac
h
iev
e
ac
cu
r
ac
ies
b
etwe
en
8
2
%
an
d
8
6
%,
wh
ile
th
e
p
r
o
p
o
s
ed
h
y
b
r
id
R
F
–
GB
T
m
o
d
el
r
ea
ch
es
9
4
.
9
3
%
o
v
er
all
ac
cu
r
ac
y
an
d
9
2
.
3
7
%
C
o
h
en
’
s
k
ap
p
a
.
As illu
s
tr
ated
in
T
ab
le
6
,
th
e
p
er
f
o
r
m
an
ce
g
a
p
clea
r
ly
h
ig
h
lig
h
ts
th
e
lim
itatio
n
s
o
f
tr
ad
itio
n
al
r
atio
-
b
ased
m
eth
o
d
s
wh
en
d
ea
lin
g
with
co
m
p
lex
o
r
b
o
r
d
er
li
n
e
f
au
lt
c
o
n
d
itio
n
s
.
T
h
ese
s
u
p
er
io
r
p
er
f
o
r
m
an
ce
s
ca
n
b
e
ex
p
lain
e
d
b
y
t
h
e
ab
ilit
y
o
f
en
s
em
b
le
m
o
d
els
to
ca
p
tu
r
e
n
o
n
lin
ea
r
an
d
cr
o
s
s
-
d
ep
en
d
e
n
cies
b
etwe
en
d
is
s
o
lv
ed
g
ases
,
wh
ich
r
atio
-
b
ased
m
eth
o
d
s
ca
n
n
o
t
r
ep
r
esen
t.
I
n
co
n
tr
ast,
co
n
v
en
tio
n
al
m
eth
o
d
s
r
ely
o
n
f
ix
ed
th
r
esh
o
ld
r
u
les
an
d
p
r
e
d
ef
in
ed
zo
n
es,
wh
ich
lim
its
th
eir
f
lex
ib
ilit
y
an
d
lead
s
to
am
b
ig
u
ity
wh
e
n
g
as
r
atio
s
f
all
n
ea
r
d
ec
is
io
n
b
o
u
n
d
ar
ies.
Ho
wev
er
,
it
is
im
p
o
r
tan
t
to
em
p
h
asize
th
at
tr
ad
itio
n
al
m
eth
o
d
s
r
em
ain
r
o
b
u
s
t,
in
ter
p
r
etab
le,
a
n
d
s
ta
n
d
ar
d
ized
,
wh
er
ea
s
AI
-
b
ased
m
o
d
els
r
e
q
u
ir
e
a
ca
r
ef
u
lly
ca
lib
r
ate
d
an
d
v
alid
a
ted
d
ataset
to
en
s
u
r
e
r
eliab
le
i
n
d
u
s
tr
ial
ap
p
licatio
n
.
T
ab
le
6
.
C
o
m
p
ar
is
o
n
o
f
t
h
e
p
e
r
f
o
r
m
a
n
ce
o
f
th
e
p
r
o
p
o
s
ed
h
y
b
r
id
R
F
–
GB
T
m
o
d
el
with
co
n
v
en
tio
n
al
m
eth
o
d
s
M
e
t
h
o
d
A
c
c
u
r
a
c
y
(
%)
F1
-
S
c
o
r
e
(
%)
C
o
h
e
n
’
s
k
a
p
p
a
(
%)
D
u
v
a
l
t
r
i
a
n
g
l
e
[
1
9
]
-
[
2
1
]
8
2
.
4
1
8
0
.
3
5
7
6
.
5
0
R
o
g
e
r
s
r
a
t
i
o
s
[
2
2
]
,
[
23]
8
4
.
7
2
8
3
.
0
0
7
8
.
9
0
I
EC
6
0
5
9
9
r
a
t
i
o
s
[
2
4
]
,
[
2
5
]
8
6
.
1
1
8
5
.
0
2
8
1
.
2
0
P
r
o
p
o
se
d
(
R
F
–
G
B
T)
9
4
.
9
3
9
4
.
8
1
9
2
.
3
7
Evaluation Warning : The document was created with Spire.PDF for Python.