I
nte
rna
t
io
na
l J
o
urna
l o
f
E
lect
rica
l a
nd
Co
m
p
ute
r
E
ng
in
ee
ring
(
I
J
E
CE
)
Vo
l.
8
,
No
.
4
,
A
u
g
u
s
t
201
8
,
p
p
.
2
2
2
0
~
2
2
2
9
I
SS
N:
2
0
8
8
-
870
8
,
DOI
: 1
0
.
1
1
5
9
1
/
i
j
ec
e
.
v8
i
4
.
p
p
2
2
2
0
-
2229
2220
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ia
e
s
co
r
e
.
co
m/
jo
u
r
n
a
ls
/in
d
ex
.
p
h
p
/
I
JE
C
E
Ev
a
lua
tion o
f
a
Multiple
Reg
ress
i
o
n Mo
del f
o
r Noi
sy
and
M
iss
ing
Data
Cha
nin
t
o
rn
J
it
t
a
w
iriy
a
nu
k
o
o
n
A
s
su
m
p
ti
o
n
Un
iv
e
rsity
,
T
h
a
il
a
n
d
Art
icle
I
nfo
AB
ST
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
No
v
13
,
2
0
1
7
R
ev
i
s
ed
J
an
17
,
2
0
1
8
A
cc
ep
ted
A
p
r
2
,
2
0
1
8
T
h
e
sta
n
d
a
rd
d
a
t
a
c
o
ll
e
c
ti
o
n
p
r
o
b
lem
s
m
a
y
in
v
o
lv
e
n
o
ise
les
s
d
a
t
a
w
h
il
e
o
n
th
e
o
th
e
r
h
a
n
d
larg
e
o
rg
a
n
iza
ti
o
n
s
c
o
m
m
o
n
ly
e
x
p
e
rien
c
e
n
o
isy
a
n
d
m
issin
g
d
a
ta,
p
ro
b
a
b
ly
c
o
n
c
e
rn
in
g
d
a
ta
c
o
ll
e
c
ted
f
ro
m
in
d
iv
id
u
a
ls.
A
s
n
o
isy
a
n
d
m
is
sin
g
d
a
ta
w
il
l
b
e
si
g
n
if
ica
n
tl
y
w
o
rriso
m
e
f
o
r
o
c
c
a
sio
n
s
o
f
th
e
v
a
st
d
a
ta
c
o
ll
e
c
ti
o
n
th
e
n
th
e
in
v
e
stig
a
ti
o
n
o
f
d
if
fe
re
n
t
f
il
terin
g
tec
h
n
iq
u
e
s
f
o
r
b
ig
d
a
ta
e
n
v
iro
n
m
e
n
t
w
o
u
ld
b
e
re
m
a
rk
a
b
le.
A
m
u
lt
ip
le
re
g
re
ss
io
n
m
o
d
e
l
w
h
e
re
b
ig
d
a
ta
is
e
m
p
lo
y
e
d
f
o
r
e
x
p
e
ri
m
e
n
ti
n
g
w
il
l
b
e
p
re
se
n
ted
.
A
p
p
ro
x
ima
ti
o
n
f
o
r
d
a
tas
e
ts
w
it
h
n
o
isy
a
n
d
m
issin
g
d
a
ta
is
a
lso
p
ro
p
o
se
d
.
T
h
e
sta
ti
stica
l
ro
o
t
m
e
a
n
sq
u
a
re
d
e
rro
r
(RM
S
E)
a
ss
o
c
iate
d
w
it
h
c
o
rre
latio
n
c
o
e
ff
icie
n
t
(COEF
)
w
il
l
b
e
a
n
a
l
y
z
e
d
to
p
ro
v
e
th
e
a
c
c
u
ra
c
y
o
f
e
sti
m
a
to
rs.
F
in
a
ll
y
,
re
su
lt
s
p
re
d
icte
d
b
y
m
a
ss
iv
e
o
n
li
n
e
a
n
a
l
y
sis
(M
OA
)
w
i
ll
b
e
c
o
m
p
a
re
d
to
th
o
se
re
a
l
d
a
ta
c
o
ll
e
c
ted
f
ro
m
th
e
f
o
ll
o
w
in
g
d
if
fe
re
n
t
ti
m
e
.
T
h
e
se
th
e
o
re
ti
c
a
l
p
re
d
ictio
n
s
w
it
h
n
o
isy
a
n
d
m
issi
n
g
d
a
ta
e
stim
a
ti
o
n
b
y
si
m
u
latio
n
,
re
v
e
a
li
n
g
c
o
n
siste
n
c
y
w
it
h
th
e
re
a
l
d
a
ta
a
re
il
lu
stra
ted
.
De
leti
o
n
m
e
c
h
a
n
ism
(DE
L
)
o
u
t
p
e
rf
o
rm
s
w
it
h
th
e
l
o
w
e
st av
e
ra
g
e
p
e
rc
e
n
tag
e
o
f
e
rro
r.
K
ey
w
o
r
d
:
B
ig
d
ata
C
las
s
i
f
icatio
n
No
is
y
a
n
d
m
i
s
s
i
n
g
d
ata
P
er
f
o
r
m
a
n
ce
ev
al
u
atio
n
R
eg
r
es
s
io
n
m
o
d
el
R
o
o
t
m
ea
n
s
q
u
ar
e
e
r
r
o
r
Co
p
y
rig
h
t
©
2
0
1
8
In
stit
u
te o
f
A
d
v
a
n
c
e
d
E
n
g
i
n
e
e
rin
g
a
n
d
S
c
ien
c
e
.
Al
l
rig
h
ts
re
se
rv
e
d
.
C
o
r
r
e
s
p
o
nd
ing
A
uth
o
r
:
C
h
a
n
i
n
to
r
n
J
itta
w
ir
i
y
an
u
k
o
o
n
,
Ass
u
m
p
t
io
n
U
n
iv
er
s
it
y
,
T
h
ail
an
d
.
E
m
ail:
p
ct2
5
2
6
@
y
a
h
o
o
.
co
m
1.
I
NT
RO
D
UCT
I
O
N
Du
r
in
g
r
ec
en
t
y
ea
r
s
,
t
h
e
co
n
ce
p
t
o
f
b
i
g
d
ata
a
n
d
t
h
e
f
l
u
ctu
ated
ap
p
licatio
n
s
o
f
t
h
e
I
n
ter
n
et
o
f
E
v
er
y
t
h
i
n
g
(
I
o
E
)
p
er
s
e
h
av
e
b
ee
n
ch
ased
w
i
th
h
u
g
e
at
ten
ti
o
n
b
y
th
e
d
ata
s
cie
n
tis
t
s
.
B
ig
d
ata
is
a
h
o
li
s
tic,
f
r
ee
f
o
r
m
atted
an
d
ti
m
e
ev
o
lv
in
g
o
f
d
y
n
a
m
ic
in
f
o
r
m
atio
n
b
u
t
in
f
l
u
e
n
ce
th
e
q
u
ali
t
y
o
f
i
n
h
ab
itan
t
lif
e
a
n
d
to
ad
v
an
ce
an
e
n
v
ir
o
n
m
en
ta
l
s
u
s
tain
ab
le
g
r
o
w
t
h
,
d
ig
ital
ec
o
n
o
m
y
a
n
d
p
u
b
l
ic
in
u
n
in
ter
r
u
p
te
d
d
ev
elo
p
m
e
n
t.
Fo
r
in
s
ta
n
ce
,
a
s
m
ar
t
a
n
d
d
ig
itize
d
cit
y
is
cr
u
cial
to
ac
cu
m
u
lat
e
v
ast
d
ata
a
n
d
to
p
o
w
er
f
u
ll
y
ca
r
r
y
o
u
t
d
ec
is
io
n
-
m
ak
in
g
o
n
t
h
e
u
p
b
r
in
g
i
n
g
.
I
n
o
r
d
er
to
ac
co
m
p
lis
h
s
o
,
it
i
s
t
o
m
a
k
e
u
s
e
o
f
d
ata
c
u
r
atio
n
s
u
ch
a
s
clas
s
if
icatio
n
o
r
clu
s
ter
in
g
w
h
ic
h
w
il
l
co
n
s
en
t
to
s
tr
aig
h
t
f
o
r
w
ar
d
l
y
r
e
-
o
r
g
an
ize
a
n
en
o
r
m
o
u
s
d
ata
w
it
h
t
h
e
asp
ir
e
o
f
in
telli
g
e
n
tl
y
d
ata
an
al
y
tic
s
,
ev
alu
a
tio
n
,
p
r
ed
ictio
n
an
d
v
is
u
aliza
tio
n
;
f
r
o
m
u
n
s
tr
u
ct
u
r
ed
d
ata
f
o
r
m
at
t
h
r
o
u
g
h
p
r
e
-
p
r
o
ce
s
s
in
g
,
to
th
e
s
cr
u
ti
n
y
o
f
n
o
is
y
d
ata
an
d
m
is
s
in
g
d
ata.
Fo
r
th
e
im
p
r
o
v
e
m
e
n
t
o
f
th
e
I
o
E
an
d
th
e
s
m
ar
t
cit
y
[
1
]
,
w
ir
ele
s
s
s
e
n
s
o
r
[
2
]
h
as
b
ee
n
d
ep
lo
y
ed
in
o
r
d
er
to
en
s
u
r
e
r
ea
l
-
ti
m
e
s
u
p
er
v
is
i
n
g
o
f
s
ev
er
al
d
ev
ices
th
at
ca
n
en
h
an
ce
d
i
g
ital
-
a
g
e
lif
e
s
t
y
le,
en
er
g
y
-
s
av
in
g
an
d
t
h
e
q
u
al
it
y
ti
m
e
o
f
co
m
m
u
n
it
y
.
T
h
is
m
ea
s
u
r
eles
s
a
m
o
u
n
t
o
f
ti
m
e
s
er
i
es
d
ata
g
e
n
er
ated
b
y
t
h
e
s
e
n
s
o
r
d
ev
ices,
s
i
m
u
lta
n
eo
u
s
l
y
w
i
th
t
h
e
co
llected
d
at
a
f
r
o
m
o
th
er
d
ig
ital
d
ev
ices,
s
u
c
h
as
ce
ll
p
h
o
n
e
s
,
m
o
n
ito
r
in
g
s
o
f
t
w
ar
e
an
d
s
o
cial
m
ed
ia,
m
u
s
t
b
e
ap
p
r
o
p
r
i
atel
y
tr
ea
s
u
r
ed
an
d
s
u
b
s
eq
u
en
t
l
y
p
r
e
-
p
r
o
ce
s
s
ed
i
n
o
r
d
er
to
s
ec
u
r
e
th
e
i
n
s
i
g
h
t
s
.
Scr
u
tin
izin
g
g
i
g
a
n
tic
a
m
o
u
n
t
o
f
d
ata
i
s
a
b
i
g
th
r
ea
t,
th
u
s
b
ig
d
ata
cu
r
atio
n
[
3
]
f
o
r
g
ath
er
in
g
,
s
a
v
in
g
an
d
in
v
est
ig
ati
n
g
d
atasets
h
as
t
o
b
e
ass
o
ciab
le
to
ex
ec
u
te
t
h
e
m
p
r
o
p
er
ly
a
n
d
r
eso
u
r
ce
f
u
ll
y
.
I
n
p
r
ac
tice,
m
is
s
in
g
d
ata
[
4
]
a
r
is
e
as
n
o
v
alu
e
(
b
lan
k
o
r
u
n
e
x
p
ec
ted
)
h
as
b
ee
n
f
o
u
n
d
f
o
r
an
y
en
tr
ie
s
d
u
r
in
g
t
h
e
s
u
r
v
eilla
n
ce
a
s
s
h
o
w
n
i
n
F
ig
u
r
e
1
.
M
is
s
in
g
d
a
ta
ca
n
co
m
m
o
n
l
y
ar
is
e
a
n
d
h
av
e
a
co
n
s
id
er
ab
le
co
n
s
eq
u
e
n
ce
o
n
th
e
f
in
al
r
es
u
lts
w
h
ic
h
ca
n
b
e
ex
tr
ac
ted
f
r
o
m
d
ata
en
tr
y
.
I
t
is
d
u
e
to
n
o
n
r
esp
o
n
s
i
v
e:
b
lan
k
in
f
o
r
m
atio
n
is
f
illed
f
o
r
at
le
ast
o
n
e
o
r
m
a
n
y
p
lace
s
.
So
m
e
p
r
iv
ate
o
r
s
e
n
s
iti
v
e
ac
co
u
n
t
s
f
o
r
ex
a
m
p
le
a
g
e,
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:
2
0
8
8
-
8708
E
va
lu
a
tio
n
o
f a
Mu
ltip
le
R
e
g
r
ess
io
n
Mo
d
el
fo
r
N
o
is
y
a
n
d
Mis
s
in
g
Da
ta
(
C
h
a
n
in
to
r
n
Jitta
w
ir
iya
n
u
ko
o
n
)
2221
in
co
m
e,
a
n
d
etc
ar
e
lik
el
y
to
p
r
o
d
u
ce
a
q
u
esti
o
n
t
h
an
o
t
h
er
s
.
Miss
in
g
n
ess
m
a
y
ar
is
e
a
s
r
esp
o
n
d
en
ts
g
i
v
e
u
p
b
ef
o
r
e
th
e
q
u
est
io
n
n
air
e
ti
m
e
ter
m
i
n
ate
s
a
n
d
ac
q
u
ir
ed
a
n
s
w
er
s
ar
e
m
is
s
i
n
g
.
Data
n
o
r
m
all
y
ar
e
in
cr
ed
ib
l
y
m
is
s
i
n
g
in
q
u
an
t
itati
v
e
r
esear
c
h
i
n
b
u
s
i
n
e
s
s
,
e
n
g
i
n
ee
r
i
n
g
a
n
d
s
cien
ce
s
s
i
n
ce
g
o
v
er
n
m
e
n
t o
f
f
icer
s
o
p
t n
o
t to
,
o
r
f
ee
l
r
el
u
cta
n
t
to
,
f
ill
cr
itic
al/s
en
s
iti
v
e
m
ea
s
u
r
e
m
en
ts
.
Occ
asio
n
all
y
th
o
s
e
m
is
ta
k
es
ar
e
m
o
ti
v
ated
b
y
r
esear
ch
er
s
,
f
o
r
in
s
ta
n
ce
,
i
m
p
r
o
p
er
d
ata
co
llectio
n
is
co
n
d
u
cted
o
r
h
u
m
a
n
-
er
r
o
r
s
ar
e
tak
en
d
u
r
in
g
d
ata
en
tr
y
p
h
ase.
T
h
ese
m
is
s
i
n
g
n
es
s
t
h
u
s
s
i
m
p
l
y
cr
ea
te
d
iv
er
s
e
ca
teg
o
r
ies,
m
is
s
i
n
g
b
y
c
h
an
ce
o
r
in
ten
tio
n
al
m
i
s
s
i
n
g
.
C
o
n
s
eq
u
en
tl
y
n
o
is
y
d
ata
t
h
at
d
etec
ted
a
m
o
n
g
th
e
n
a
v
i
g
atio
n
m
e
n
u
,
ad
v
er
tis
i
n
g
b
an
n
er
an
d
o
th
er
in
f
o
r
m
atio
n
co
n
ten
t
o
f
t
h
e
w
eb
d
o
cu
m
e
n
t
in
f
lu
e
n
ce
s
ad
v
er
s
e
l
y
t
h
ep
er
f
o
r
m
an
ce
o
f
m
o
b
ile
ap
p
licatio
n
s
th
at
in
v
o
lv
e
s
w
it
h
th
ec
o
n
te
n
t
a
s
s
u
c
h
[
5
]
.
I
n
ad
d
itio
n
,
S
h
ar
m
a
a
n
d
B
h
at
ia
[
6
]
ex
p
an
d
ed
a
p
ag
e
r
ep
lace
m
en
t
a
lg
o
r
ith
m
i
n
o
r
d
er
to
s
ep
ar
ate
n
o
is
y
d
ata
f
r
o
m
w
eb
d
o
cu
m
e
n
t
.
C
h
ae
et
a
l
.
[
7
]
m
en
tio
n
it
is
co
m
p
u
l
s
o
r
y
t
h
at
b
ig
d
ata
an
a
l
y
t
ics
i
n
s
u
p
p
l
y
ch
ai
n
m
a
n
ag
e
m
e
n
t
(
S
C
M)
b
e
co
llecti
v
e
w
it
h
SC
M
o
b
j
ec
tiv
es
to
ad
v
an
ce
w
o
r
k
i
n
g
p
er
f
o
r
m
a
n
ce
a
n
d
escalat
e
th
e
v
al
u
e
o
f
an
a
l
y
t
ics.
Ho
w
e
v
er
,
th
e
y
h
a
v
e
n
o
t b
ee
n
lo
o
k
in
g
at
p
r
ac
tical
is
s
u
e
o
f
m
is
s
i
n
g
n
e
s
s
.
No
is
y
d
ata
is
d
escr
ib
ed
as
w
o
r
th
les
s
d
ata.
T
h
e
ter
m
i
s
ca
l
led
as a
n
alter
n
ati
v
e
e
x
p
r
ess
io
n
f
o
r
cr
o
o
k
ed
d
ata
as
d
ep
icted
in
F
ig
u
r
e
2
.
No
n
eth
ele
s
s
,
t
h
e
m
ea
n
in
g
h
a
s
in
clu
d
ed
an
y
i
n
co
m
p
r
eh
e
n
s
iv
e
d
ata
f
o
r
in
s
tan
ce
u
n
s
tr
u
ct
u
r
ed
f
o
r
m
at
o
f
d
ata.
An
y
u
n
r
ea
d
ab
le
d
ata
w
h
ich
h
as
b
ee
n
d
etec
ted
b
y
th
e
m
ac
h
i
n
e
w
ill
d
e
v
elo
p
an
d
ca
n
b
e
d
ef
in
ed
a
s
n
o
is
e.
S
h
a
b
ir
an
d
P
ad
m
a
[
8
]
p
r
esen
ted
a
d
en
o
is
e
p
r
o
ce
d
u
r
e
to
i
m
p
r
o
v
e
th
e
q
u
alit
y
o
f
o
r
ig
in
al
i
m
ag
e.
No
is
y
d
ata
is
w
o
r
s
e
n
i
n
g
o
f
d
ata
co
llectio
n
ca
u
s
ed
b
y
e
x
ter
n
al
h
az
ar
d
s
.
T
h
ese
n
o
is
e
i
n
cl
u
d
e
n
o
t
o
n
l
y
i
n
ter
n
al
p
r
o
b
le
m
s
s
u
ch
as
s
o
f
t
w
ar
e
o
r
h
ar
d
w
ar
e
in
co
m
p
atib
ilit
y
o
r
v
ir
u
s
e
s
,
s
y
s
te
m
m
alf
u
n
ctio
n
,
f
ail
u
r
es,
o
r
f
la
w
s
,
b
u
t
also
en
v
ir
o
n
m
en
tal
h
az
ar
d
s
s
u
c
h
as
d
u
s
t,
m
o
is
t,
e
x
tr
e
m
e
te
m
p
er
atu
r
es,
b
lack
-
o
u
ts
,
an
d
w
ater
.
No
is
y
d
ata
o
n
th
e
o
th
e
r
h
an
d
r
ed
u
n
d
an
tl
y
r
eq
u
ir
es
t
h
e
ex
tr
ao
r
d
in
ar
y
a
m
o
u
n
t
o
f
s
a
v
in
g
s
p
ac
e
an
d
ca
n
u
n
f
a
v
o
r
ab
l
y
u
p
s
et
t
h
e
o
u
tco
m
es o
f
d
ata
an
al
y
tics
.
T
h
u
s
an
al
y
s
is
ca
n
o
v
er
co
m
e
t
h
is
p
r
o
b
le
m
b
y
e
m
p
lo
y
i
n
g
d
at
a
co
llected
p
r
ev
io
u
s
l
y
(
h
is
to
r
i
ca
l
d
ata)
to
f
ilter
o
u
t
n
o
is
y
d
ata
an
d
ea
s
e
d
ata
cu
r
atio
n
.
Mu
c
h
o
f
n
o
is
y
d
a
ta
ca
n
af
f
ec
t
f
ail
u
r
es
in
h
ar
d
war
e
p
r
o
ce
s
s
in
g
an
d
ac
cu
r
ac
y
.
Mo
r
eo
v
er
,
t
y
p
o
s
,
s
lan
g
,
m
i
s
s
p
elli
n
g
,
ca
r
eless
an
d
o
th
er
ab
b
r
ev
iat
i
o
n
s
ca
n
o
b
s
tr
u
ct
m
ac
h
i
n
e
lear
n
in
g
.
C
o
r
r
u
p
t
d
ata
is
a
r
e
alis
tic
tr
o
u
b
le,
i
n
d
u
ce
d
eit
h
er
b
ec
au
s
e
o
f
d
e
f
ec
ti
v
e
d
ata
s
o
u
r
ce
s
o
r
d
u
r
in
g
d
ata
b
r
o
ad
ca
s
t
(
tr
av
er
s
in
g
)
.
No
is
e
is
ab
le
to
s
er
io
u
s
l
y
m
es
s
u
p
m
ac
h
in
e
lear
n
i
n
g
p
r
o
ce
s
s
o
f
c
o
llected
d
ata.
I
t
is
a
m
u
c
h
m
o
r
e
r
i
g
o
r
o
u
s
tr
o
u
b
le
i
n
ca
s
e
o
f
d
ata
s
tr
ea
m
s
a
s
it
co
n
n
ec
t
s
w
i
th
co
n
ce
p
t
d
r
if
t.
I
f
a
g
r
ee
d
y
al
g
o
r
ith
m
is
co
n
ce
r
n
ed
to
co
n
ce
p
t
d
r
if
t,
it
m
a
y
q
u
ali
f
y
n
o
is
e
b
y
er
r
o
n
eo
u
s
l
y
p
ictu
r
i
n
g
it
a
s
d
ata
f
r
o
m
a
f
r
esh
co
n
ce
p
t.
I
f
i
t
is
to
b
e
to
o
s
tr
ict
to
n
o
is
e,
it
m
a
y
h
a
v
e
to
ig
n
o
r
e
d
r
if
ts
th
e
n
f
in
e
-
t
u
n
e.
B
esid
es,
th
e
co
m
p
u
tatio
n
a
l
co
m
p
lex
it
y
o
f
th
e
K
-
m
ea
n
s
alg
o
r
it
h
m
w
it
h
d
atasets
h
as b
ee
n
e
v
al
u
ated
i
n
[
9
]
.
Fig
u
r
e
1
.
E
x
am
p
le
o
f
m
is
s
in
g
d
at
a
(
b
o
l
d
o
r
an
g
e
)
T
h
e
ai
m
o
f
t
h
is
p
ap
er
is
to
ev
alu
a
te
t
h
e
ac
c
u
r
ac
y
o
f
m
u
l
tip
le
r
eg
r
ess
io
n
a
n
al
y
s
i
s
f
o
r
n
o
is
y
a
n
d
m
is
s
i
n
g
d
ata
en
v
ir
o
n
m
e
n
t
u
s
i
n
g
MO
A
[
10
]
s
i
m
u
latio
n
.
B
o
th
n
o
is
e
a
n
d
m
i
s
s
i
n
g
n
e
s
s
w
il
l
b
e
ex
p
er
im
e
n
tall
y
tak
en
in
to
co
n
s
id
er
atio
n
f
o
r
p
r
ac
tical
p
o
in
t
o
f
v
ie
w
.
Firstl
y
,
th
e
n
o
is
y
d
ata
w
ill
b
e
w
ee
d
ed
o
u
t
in
o
r
d
er
to
av
o
id
co
m
p
li
ca
tio
n
i
n
p
r
o
ce
s
s
in
g
.
Seco
n
d
l
y
,
a
n
esti
m
ato
r
will
ir
o
n
o
u
t
m
is
s
in
g
d
ata
th
e
n
m
u
ltip
le
r
eg
r
es
s
io
n
r
esu
lt
s
f
r
o
m
s
i
m
u
latio
n
ar
e
co
l
lecte
d
f
o
r
v
alid
atio
n
o
f
th
e
e
s
t
i
m
atio
n
m
e
th
o
d
.
L
a
s
tl
y
,
p
r
ed
i
ctio
n
d
ata
b
ased
o
n
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8708
I
n
t J
E
l
ec
&
C
o
m
p
E
n
g
,
Vo
l.
8
,
No
.
4
,
A
u
g
u
s
t 2
0
1
8
:
2
2
2
0
–
2
2
2
9
2222
m
u
ltip
le
r
eg
r
e
s
s
io
n
eq
u
at
io
n
w
il
l
b
e
co
m
p
ar
ed
t
o
f
o
llo
w
in
g
r
ea
l
d
ata.
T
h
e
ac
c
u
r
ac
y
o
f
n
o
is
e
f
ilter
i
n
g
a
n
d
n
o
n
-
f
ilter
i
n
g
tec
h
n
iq
u
es t
h
e
n
w
il
l b
e
d
is
cu
s
s
ed
af
ter
co
m
p
ar
is
o
n
.
Fig
u
r
e
2
.
E
x
a
m
p
le
o
f
n
o
is
y
d
a
ta
2.
RE
L
AT
E
D
WO
RK
I
n
o
r
d
er
to
f
r
esh
th
e
te
x
t
d
o
cu
m
e
n
t
b
y
r
e
m
o
v
i
n
g
th
e
n
o
is
y
d
ata
o
u
t
h
as
b
ee
n
d
is
c
u
s
s
ed
in
ten
s
i
v
el
y
u
s
i
n
g
v
ar
io
u
s
tec
h
n
iq
u
e
s
.
B
ar
-
Yo
s
s
e
f
e
t
a
l
.
[
1
1
]
ap
p
lied
an
ap
p
r
o
ac
h
b
ased
u
p
o
n
d
o
cu
m
en
t
o
b
j
ec
t
m
o
d
el’
s
tr
ee
o
f
t
h
e
w
eb
co
n
te
n
t.
An
o
t
h
er
m
eth
o
d
w
h
ic
h
e
x
tr
ac
ts
n
o
is
e
i
n
o
r
d
er
to
ad
v
a
n
ce
d
ata
m
i
n
in
g
o
u
tco
m
es
h
a
s
b
ee
n
p
r
o
p
o
s
ed
b
y
[
1
2
]
.
T
h
e
y
f
o
r
m
er
l
y
e
m
p
lo
y
a
s
tr
u
ct
u
r
al
tr
ee
w
h
ich
is
an
a
lo
g
o
u
s
to
t
h
e
tr
ee
s
tr
u
c
tu
r
e
o
f
d
o
cu
m
en
t
o
b
j
ec
t
m
o
d
el.
Af
te
r
th
at
th
e
n
o
is
y
d
ata
w
i
ll
b
e
d
is
co
v
er
ed
b
y
a
n
e
v
al
u
atio
n
p
r
o
ce
s
s
o
f
co
n
ten
ts
f
o
u
n
d
in
s
tr
u
c
tu
r
al
tr
ee
.
Ho
w
e
v
er
,
co
n
s
tr
u
c
tio
n
o
f
th
e
p
r
ef
er
ab
le
tr
ee
a
s
s
u
ch
w
ill
b
e
ti
m
e
-
co
n
s
u
m
p
tio
n
.
Deb
n
ath
et
a
l
.
[
1
3
]
h
av
e
r
ec
o
m
m
en
d
ed
a
tech
n
iq
u
e
co
m
p
ar
ab
le
to
m
et
h
o
d
w
r
it
ten
i
n
[
1
2
]
,
o
n
th
e
o
th
er
h
a
n
d
r
ath
er
ch
o
s
e
n
d
ata
b
lo
ck
s
,
w
h
i
ch
ar
e
n
o
n
tr
i
v
ial
b
u
t
ex
ce
ed
a
s
et
th
r
e
s
h
o
ld
.
T
h
en
in
d
i
v
id
u
a
l
d
ata
b
lo
ck
w
ill
b
e
p
r
o
j
ec
t
ed
.
No
t
all
co
llected
d
ata
s
u
cc
ee
d
s
t
h
e
s
a
m
e
m
i
s
t
ak
en
l
y
d
e
f
ec
ti
v
e
p
atter
n
.
A
m
et
h
o
d
w
h
ic
h
ca
n
p
in
p
o
in
t
an
d
r
e
m
o
v
e
p
er
tin
e
n
t
co
n
ten
t
f
r
o
m
w
eb
co
n
te
x
t
s
h
a
s
b
ee
n
in
tr
o
d
u
ce
d
b
y
[
1
4
]
.
T
h
e
m
e
th
o
d
s
p
ec
if
ie
s
s
ig
n
i
f
ica
n
ce
as
tex
t
s
w
h
ic
h
m
o
r
e
in
d
ep
en
d
en
tl
y
in
ter
p
r
et
ab
le
th
an
th
e
i
m
a
g
e.
B
u
t,
th
e
m
et
h
o
d
f
ac
ilit
ates
o
n
l
y
o
n
w
eb
co
n
tex
t
s
.
I
t
s
h
o
u
ld
b
e
an
u
n
co
m
p
licated
a
lg
o
r
it
h
m
w
h
ic
h
ca
n
n
eu
tr
all
y
ex
tr
ac
t
n
o
is
y
d
ata
f
r
o
m
t
h
e
w
eb
.
A
n
o
is
e
r
ed
u
c
tio
n
al
g
o
r
ith
m
w
ith
t
h
r
ee
p
h
ase
s
h
a
s
b
ee
n
p
r
esen
ted
in
[
1
5
]
.
T
h
e
1
st
p
h
ase
as
s
p
ec
i
f
ied
in
t
h
is
al
g
o
r
ith
m
co
n
v
er
t
s
a
w
eb
d
o
cu
m
en
t
to
tab
le
f
o
r
m
a
t
co
n
t
ain
i
n
g
o
f
x
in
s
ta
n
ce
s
a
n
d
y
attr
ib
u
te
s
,
t
h
e
n
t
h
e
e
s
s
e
n
tial
d
ata
ca
n
b
e
a
d
d
ed
in
to
th
e
tab
le
i
f
n
ec
es
s
a
r
y
.
T
h
e
2
nd
a
n
d
3
rd
p
h
ase
w
ill
s
o
lel
y
w
ee
d
o
u
t
n
o
is
y
d
ata
u
s
in
g
f
ilter
in
g
tec
h
n
i
q
u
es.
Oth
er
o
p
er
atio
n
al
ag
g
r
eg
ated
m
et
h
o
d
s
s
p
li
t
d
ata
s
tr
ea
m
s
in
to
f
i
x
ed
b
lo
ck
s
a
n
d
m
ac
h
i
n
e
w
ill
lear
n
an
a
g
g
r
e
g
atio
n
f
r
o
m
in
d
i
v
id
u
al
b
l
o
ck
s
.
As
s
o
o
n
a
s
a
f
u
n
d
a
m
en
ta
l
m
o
d
el
i
s
b
u
ilt,
i
t
w
ill
n
ev
er
a
m
en
d
n
e
w
s
tr
ea
m
s
.
I
n
g
e
n
er
al
t
h
er
e
ar
e
t
w
o
v
o
tin
g
ca
te
g
o
r
ies
:
u
n
i
f
o
r
m
an
d
w
eig
h
ted
.
T
h
ese
t
w
o
t
y
p
e
s
ar
e
n
o
t
as
s
o
ciate
d
to
o
u
r
ap
p
r
o
ac
h
as
o
u
r
p
r
o
p
o
s
ed
m
et
h
o
d
w
ill
co
n
s
tr
u
ct
an
ag
g
r
eg
a
tio
n
f
r
o
m
t
h
ese
s
eq
u
e
n
tia
l
d
iv
id
ed
b
lo
ck
s
.
B
u
t
w
h
at
i
s
o
m
itted
f
r
o
m
th
e
ab
o
v
e
t
w
o
ap
p
r
o
ac
h
es
is
an
a
n
al
y
tical
to
o
l
h
an
d
li
n
g
n
o
is
y
d
at
a.
W
h
ile
th
er
e
ar
e
s
o
m
e
al
g
o
r
ith
m
s
f
o
r
n
o
is
e
r
ec
o
g
n
itio
n
,
s
o
ca
lled
an
o
m
al
y
d
etec
tio
n
,
n
o
is
e
eli
m
in
a
tio
n
cr
a
f
ts
a
c
o
n
s
id
er
ab
le
b
r
ea
ch
b
et
w
ee
n
d
ata
s
tr
ea
m
an
d
t
h
e
ab
o
v
e
m
en
t
io
n
ed
ap
p
r
o
ac
h
es.
Fu
r
t
h
er
m
o
r
e,
th
e
is
s
u
e
o
f
ac
q
u
ir
in
g
a
n
o
is
y
d
ata
w
i
ll b
e
ad
d
r
ess
ed
.
Ou
r
s
t
u
d
y
t
h
e
n
is
u
n
lik
e
t
w
o
a
p
p
r
o
ac
h
es
as
s
tated
ab
o
v
e,
f
ir
s
tl
y
,
a
n
o
m
al
y
d
etec
tio
n
w
ill
b
e
clen
ch
ed
u
p
in
to
th
e
m
ac
h
i
n
e
lear
n
in
g
p
r
o
ce
s
s
f
o
r
t
h
e
r
ea
s
o
n
th
a
t
co
n
ce
p
t
d
r
if
t
i
s
a
f
r
ac
tio
n
o
f
o
u
tlier
d
etec
tio
n
.
Seco
n
d
l
y
,
t
h
e
d
is
tan
ce
v
ec
to
r
ca
n
b
e
ea
s
ily
d
r
a
w
n
b
y
t
h
e
class
i
f
ier
,
r
ath
er
f
r
o
m
a
f
o
r
m
u
la
s
p
ec
if
ied
b
y
d
atasets
p
er
s
e.
A
s
a
m
atter
o
f
f
ac
t,
th
e
a
n
o
m
al
y
d
etec
tio
n
a
n
d
ad
ap
tiv
e
m
ac
h
i
n
e
lear
n
in
g
h
en
ce
r
ec
ip
r
o
ca
ll
y
s
u
p
p
o
r
t
o
n
e
an
o
th
er
.
I
n
g
e
n
er
al,
an
ac
cu
r
ate
ad
ap
tiv
e
m
o
d
el
n
o
u
r
is
h
es
to
d
is
co
v
er
th
e
an
o
m
alie
s
.
A
lter
n
ati
v
el
y
,
b
y
p
r
o
p
er
ly
f
in
d
in
g
an
d
r
e
m
o
v
in
g
th
e
an
o
m
a
lies
at
ea
r
lier
s
tag
e,
a
f
u
r
th
er
ex
ac
t
m
o
d
el
ca
n
b
e
ex
ec
u
ted
.
A
d
ap
tiv
e
lear
n
i
n
g
m
o
d
el
w
it
h
r
ef
er
e
n
ce
to
v
ig
o
r
an
d
r
ev
is
io
n
,
co
r
r
esp
o
n
d
in
g
l
y
w
i
ll
b
e
illu
s
tr
ated
.
Mo
d
el
m
ap
p
in
g
a
n
d
ca
lcu
latio
n
w
ill b
e
p
r
o
v
id
ed
as
w
ell.
Af
t
er
w
ar
d
in
v
esti
g
atio
n
al
r
es
u
lts
w
il
l b
e
lis
ted
.
3.
NO
I
SE
A
ND
M
I
SS
I
N
G
DA
T
AS
E
T
S
I
n
th
is
s
ec
tio
n
,
ch
ar
ac
ter
is
tic
s
o
f
n
o
is
e
w
ill
b
e
d
escr
ib
ed
,
at
th
e
s
a
m
e
ti
m
e
d
atasets
w
h
ic
h
ar
e
in
cl
u
s
i
v
e
o
f
n
o
is
e
ar
e
o
u
tli
n
ed
an
d
th
e
co
m
p
r
eh
e
n
s
i
v
e
d
is
c
u
s
s
io
n
is
g
i
v
en
.
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:
2
0
8
8
-
8708
E
va
lu
a
tio
n
o
f a
Mu
ltip
le
R
e
g
r
ess
io
n
Mo
d
el
fo
r
N
o
is
y
a
n
d
Mis
s
in
g
Da
ta
(
C
h
a
n
in
to
r
n
Jitta
w
ir
iya
n
u
ko
o
n
)
2223
3
.
1
.
S
elf
-
g
ener
a
t
ed
no
is
e
Ou
r
ap
p
r
o
ac
h
d
escr
ib
e
d
in
th
is
p
ap
er
w
ill
allo
w
MO
A
s
i
m
u
latio
n
to
r
ea
d
a
n
o
is
y
d
at
aset.
MO
A
to
ler
ates
co
m
p
ar
is
o
n
o
f
s
in
g
le
alg
o
r
ith
m
o
n
d
ataset
s
w
it
h
d
i
s
s
i
m
ilar
n
o
is
e
r
ate
s
.
T
h
e
p
r
o
c
ed
u
r
e
is
cr
ea
ted
b
y
n
u
m
er
o
u
s
id
ea
s
.
A
d
atase
t
r
ea
d
er
in
itiall
y
w
il
l
ch
o
o
s
e
s
m
all
f
r
a
g
m
e
n
t
s
w
it
h
s
p
ec
i
f
ic
v
alu
e
f
r
o
m
d
ataset
.
Seco
n
d
l
y
,
a
m
p
le
f
i
g
u
r
es
f
o
r
t
h
e
s
elec
ted
s
a
m
p
le
w
ill
b
e
ab
an
d
o
n
ed
if
it a
p
p
ea
r
s
v
er
y
d
o
u
b
t
f
u
l.
A
s
ea
ch
lea
f
i
n
d
ec
is
io
n
tr
ee
is
d
esig
n
ed
,
all
f
ig
u
r
e
s
f
r
o
m
ea
ch
attr
ib
u
te
ar
e
m
ea
s
u
r
ed
as
ap
p
lican
ts
f
o
r
f
r
a
g
m
e
n
ti
n
g
.
T
h
r
o
u
g
h
ev
er
y
r
o
u
n
d
o
f
ca
lcu
latio
n
th
a
t
d
o
es
n
o
t
d
ec
id
e
to
f
r
ag
m
e
n
ti
ze
,
attr
ib
u
tes
ar
e
m
ar
k
ed
to
b
e
u
n
f
o
r
tu
n
ate
if
t
h
eir
v
alu
e
s
ar
e
les
s
t
h
a
n
th
e
v
al
u
e
f
r
o
m
to
p
attr
ib
u
te
w
h
ich
is
g
r
ea
ter
th
a
n
t
h
e
b
o
u
n
d
.
R
e
g
a
r
d
in
g
to
t
h
e
b
o
u
n
d
,
cu
r
r
en
t
attr
ib
u
te
s
ar
e
i
m
p
r
o
b
ab
le
to
b
e
ch
o
s
en
in
t
h
e
d
ec
is
io
n
tr
ee
,
th
er
ef
o
r
e
th
e
r
ef
er
en
ce
to
th
is
in
f
o
r
m
a
tio
n
is
r
ej
ec
ted
f
r
o
m
th
at
ca
lcu
l
atio
n
p
o
in
t
o
n
w
ar
d
.
T
h
e
t
w
o
id
ea
s
ar
e
in
ter
co
n
n
ec
ted
,
r
ep
ea
tin
g
u
n
til
t
h
e
r
ep
lace
ab
le
v
a
lu
e
is
s
et.
T
h
is
ap
p
r
o
ac
h
ca
n
f
u
n
ctio
n
a
s
a
f
o
u
n
d
atio
n
f
o
r
ca
lcu
lat
in
g
a
r
an
g
e
o
f
p
r
ed
ictab
le
v
alu
e
s
f
o
r
a
n
o
i
s
y
d
ataset.
MO
A
s
i
m
u
latio
n
ca
n
r
an
d
o
m
l
y
au
g
m
en
t
n
o
is
e
to
d
atase
ts
.
P
er
tu
r
b
atio
n
a
m
o
u
n
t
i
s
p
r
esen
ted
to
s
tatis
t
ical
f
ig
u
r
es.
A
le
v
el
o
f
n
o
is
e
ca
n
b
e
f
a
m
iliar
ized
to
th
e
d
atasets
af
ter
p
r
o
v
o
k
in
g
.
Fo
r
d
is
ti
n
ct
attr
ib
u
te
s
,
a
p
er
tu
r
b
atio
n
p
r
o
b
ab
ilit
y
g
o
v
e
r
n
s
t
h
e
co
in
cid
e
n
ce
w
h
ic
h
a
n
y
f
i
g
u
r
e
s
ar
e
c
h
a
n
g
ed
to
o
th
e
r
s
b
u
t
th
e
o
r
ig
i
n
al
f
i
g
u
r
e.
I
n
t
h
e
ca
s
e
o
f
s
tatis
tical
attr
ib
u
te
s
,
a
lev
el
o
f
r
an
d
o
m
p
er
tu
r
b
ati
o
n
i
s
r
ec
k
o
n
ed
to
all
f
i
g
u
r
es,
r
an
d
o
m
ized
f
r
o
m
a
Ga
u
s
s
ian
d
is
tr
ib
u
tio
n
f
u
n
ctio
n
w
it
h
th
e
id
en
tical
s
ta
n
d
ar
d
d
ev
iatio
n
as
o
f
th
e
o
r
ig
i
n
al
f
i
g
u
r
es
t
i
m
ed
b
y
a
p
r
o
b
ab
ilit
y
o
f
p
er
tu
r
b
atio
n
.
Fo
r
ex
a
m
p
le,
th
e
alg
o
r
it
h
m
m
a
y
au
g
m
en
t
1
0
%
p
er
tu
r
b
atio
n
to
th
e
d
ec
is
io
n
tr
ee
d
ataset.
I
t
i
s
ai
m
ed
t
h
at
ex
p
er
i
m
en
t
w
it
h
n
o
is
eless
a
n
d
n
o
i
s
y
d
ata
ca
n
co
n
tr
ib
u
te
p
er
ce
p
tio
n
to
h
o
w
s
m
ar
t t
h
e
alg
o
r
it
h
m
s
c
an
s
u
cc
ee
d
p
er
tu
r
b
atio
n
.
T
o
q
u
an
tify
n
o
is
e
i
n
th
e
d
ataset
is
n
o
w
co
n
s
id
er
ed
.
On
l
y
t
h
e
ca
s
e
in
w
h
ich
a
b
o
u
n
d
o
n
th
e
n
o
is
e
ex
is
t
s
an
d
t
h
e
ca
s
e
w
h
er
e
th
e
n
o
is
e
is
r
a
n
d
o
m
.
I
n
t
h
e
f
ir
s
t
o
n
e,
o
p
ti
m
izatio
n
i
s
g
u
ar
an
te
ed
o
n
th
e
m
ac
h
i
n
e
lear
n
ed
s
i
m
u
latio
n
.
I
n
th
e
latt
er
o
n
e,
th
e
m
ac
h
i
n
e
lear
n
i
n
g
w
il
l
ass
o
ciate
s
y
n
th
e
tic
d
atase
ts
an
d
tr
ain
i
n
g
s
et.
T
o
s
y
n
t
h
e
s
ize
t
h
e
tar
g
eted
d
a
tasets
w
i
th
n
o
i
s
e,
a
s
y
n
t
h
esiz
er
f
o
r
b
it
lev
el
w
ill
v
i
g
o
r
o
u
s
l
y
in
s
er
t
s
o
m
e
f
a
ls
e
u
n
t
il
t
h
e
p
r
ef
er
r
ed
d
atasets
ar
e
ac
h
iev
ed
.
I
n
ca
s
e
o
f
h
e
f
t
y
d
ataset
s
,
a
n
u
m
er
ical
s
y
n
t
h
esizer
w
h
ic
h
ca
n
ca
lcu
late
p
r
o
b
ab
ilis
tic
m
o
d
els
f
o
r
a
d
ataset
in
r
ep
o
s
ito
r
y
th
e
n
ap
p
r
o
x
i
m
ate
s
t
h
e
p
r
ef
e
r
r
ed
d
ataset
b
ased
u
p
o
n
th
e
co
m
p
u
tatio
n
.
3
.
2
.
E
x
peri
m
e
nta
l da
t
a
s
et
s
E
li
m
i
n
ati
n
g
s
u
b
s
ta
n
ce
s
w
h
ic
h
ar
e
n
o
is
e
is
a
v
ital
o
b
j
ec
tiv
e
o
f
d
ata
c
u
r
atio
n
(
b
o
th
f
il
ter
in
g
a
n
d
o
u
tlier
)
s
in
ce
n
o
is
e
g
ar
b
les
an
d
o
b
s
tr
u
cts
d
ata
an
a
l
y
t
ics.
Su
r
v
iv
i
n
g
f
ilter
in
g
tec
h
n
iq
u
es
e
m
p
h
a
s
ize
o
n
eli
m
i
n
ati
n
g
n
o
is
e
w
h
ic
h
is
th
e
cr
ea
tio
n
o
f
lo
w
-
lev
e
l
d
ata
er
r
o
r
s
d
ev
elo
p
ed
b
y
a
d
ef
icie
n
t
d
ata
co
llectio
n
,
b
u
t
d
ata
s
u
b
s
tan
ce
s
w
h
ich
ar
e
in
a
p
p
r
o
p
r
iate
o
r
d
im
l
y
r
elate
d
ca
n
ea
s
il
y
f
r
u
s
tr
ate
t
h
e
an
al
y
tics
.
Fo
r
ex
a
m
p
le,
n
o
is
e
ca
n
d
ir
ec
t
to
b
ia
s
ed
v
ar
i
ab
les
(
n
eg
ati
v
e
co
n
s
eq
u
e
n
ce
)
,
r
es
u
lt
in
g
d
ata
s
c
ien
t
is
t
to
co
n
s
id
er
t
h
at
a
n
a
s
s
o
ciatio
n
o
f
an
y
attr
ib
u
tes
o
cc
u
r
s
(
f
a
u
lt
co
n
clu
s
io
n
)
t
h
o
u
g
h
in
f
ac
t
i
t
m
a
y
n
o
t
b
e
(
a
t
y
p
e
o
n
e
er
r
o
r
)
.
Hen
ce
,
if
i
t
is
to
en
r
ich
d
ata
a
n
al
y
tics
as
f
ar
as
ac
h
iev
ab
le,
t
h
ese
s
u
b
s
tan
ce
s
m
u
s
t
b
e
d
etec
ted
as
n
o
is
e,
r
e
g
ar
d
in
g
to
th
e
b
as
ic
an
al
y
tics
.
T
h
er
ef
o
r
e,
it
is
ess
e
n
tial
f
o
r
n
o
is
e
r
e
m
o
v
a
l
tech
n
i
q
u
es
in
o
r
d
er
to
eli
m
in
ate
a
n
y
t
y
p
es
o
f
n
o
i
s
e.
Du
e
to
b
ig
p
o
r
tio
n
s
o
f
n
o
is
e
d
u
r
i
n
g
d
ata
co
llectio
n
p
r
o
ce
s
s
,
a
n
y
tech
n
iq
u
es
h
a
v
e
to
ab
an
d
o
n
a
f
r
ac
tio
n
o
f
d
ata.
T
h
is
r
esear
ch
in
v
e
s
ti
g
ates
3
ty
p
e
s
p
r
o
p
o
s
ed
f
o
r
n
o
is
e
(
N)
,
m
is
s
in
g
d
ata
(
MD
)
an
d
th
e
in
te
g
r
atio
n
o
f
b
o
th
(
NM
D)
.
T
h
r
ee
p
r
o
p
o
s
ed
tr
ea
t
m
en
t
tech
n
iq
u
e
s
in
cl
u
d
e
L
is
t
w
i
s
e
Deltio
n
(
DE
L
)
,
Sin
g
le
Ass
er
tio
n
Me
c
h
an
is
m
(
SA
M)
,
a
n
d
R
a
n
d
o
m
Me
th
o
d
(
R
AM
)
to
i
m
p
r
o
v
e
d
ata
an
al
y
tic
s
i
n
t
h
e
o
cc
u
r
r
en
ce
o
f
ex
tr
ao
r
d
in
ar
y
n
o
is
e
lev
els.
T
h
r
ee
in
v
e
s
tig
a
tio
n
s
an
d
t
w
o
tec
h
n
iq
u
es
ar
e
b
ased
u
p
o
n
m
u
ltip
le
r
e
g
r
ess
io
n
m
o
d
el
w
it
h
f
i
v
e
d
if
f
er
e
n
t
d
atasets
.
T
h
ese
ex
p
er
i
m
en
ts
ar
e
esti
m
ated
in
ter
m
s
o
f
th
eir
i
m
p
ac
t
o
n
th
e
s
u
cc
e
s
s
i
v
e
d
ata
an
al
y
tics
,
ex
p
licit
l
y
,
th
e
f
o
llo
w
in
g
y
ea
r
d
ata
co
llect
io
n
w
ill
b
e
e
m
p
lo
y
ed
to
co
m
p
ar
e
to
th
o
s
e
e
s
ti
m
atio
n
s
f
r
o
m
m
u
ltip
le
r
eg
r
e
s
s
io
n
an
al
y
s
is
.
4.
SI
M
UL
AT
I
O
N
R
E
S
UL
T
S
AND
ANA
L
YS
I
S
T
h
e
o
p
en
-
s
o
u
r
ce
a
n
d
ac
ce
p
tab
le
to
o
l
MO
A
(
R
elea
s
e
2
0
1
7
.
0
6
)
[
10
]
w
ill
b
e
e
m
p
lo
y
ed
f
o
r
th
e
an
al
y
s
is
.
T
w
o
d
is
s
i
m
ilar
d
ata
s
ets
h
a
v
e
b
ee
n
u
s
ed
an
d
th
e
p
er
f
o
r
m
a
n
ce
ev
al
u
atio
n
o
f
a
m
u
ltip
le
r
eg
r
ess
io
n
m
o
d
el
f
o
r
n
o
is
y
an
d
m
i
s
s
i
n
g
d
ata
h
a
s
b
ee
n
f
i
g
u
r
ed
o
u
t.
T
h
e
co
m
p
u
tatio
n
h
as
b
ee
n
e
x
ec
u
ted
o
n
a
n
A
ce
r
W
in
d
o
w
s
8
w
it
h
I
n
tel®
C
o
r
e
™
i5
C
P
U,
1
.
6
0
GHz
P
r
o
ce
s
s
o
r
an
d
8
GB
R
A
M
o
n
b
o
ar
d
.
T
h
e
d
atasets
h
av
e
b
ee
n
s
elec
ted
in
o
r
d
er
th
at
th
e
y
ar
e
d
if
f
er
e
n
t i
n
n
u
m
b
er
o
f
at
tr
ib
u
tes,
in
s
ta
n
ce
s
a
n
d
s
ize.
T
h
e
p
u
b
lic
h
ea
lth
p
ar
a
m
eter
s
ar
e
co
m
p
o
s
ed
o
f
m
o
n
th
l
y
s
alar
y
(
S
AL
)
,
lo
s
s
r
ate
p
er
th
o
u
s
a
n
d
p
o
p
u
latio
n
s
(
L
OS)
,
n
u
m
b
er
o
f
m
ed
ical
d
o
cto
r
s
p
er
h
u
n
d
r
ed
th
o
u
s
an
d
p
o
p
u
latio
n
(
DO
C
)
,
n
u
m
b
er
o
f
h
o
s
p
itals
p
er
h
u
n
d
r
ed
th
o
u
s
a
n
d
p
o
p
u
latio
n
(
HOS)
an
d
p
o
p
u
latio
n
d
e
n
s
it
y
p
er
s
q
u
ar
e
k
ilo
m
eter
s
(
DE
N)
.
P
u
b
lic
h
ea
lth
d
ataset
w
a
s
p
r
o
v
in
cial
l
y
co
ll
ec
ted
b
y
i
n
f
o
r
m
atio
n
tec
h
n
o
l
o
g
is
t
w
h
o
h
ad
b
ee
n
w
o
r
k
in
g
clo
s
el
y
in
ter
m
s
o
f
p
u
b
lic
h
ea
lt
h
p
o
lic
y
an
d
p
r
o
m
o
tio
n
(
P
HP
P
)
.
C
o
llected
d
a
ta
ar
e
s
u
b
j
ec
ted
to
h
elp
m
an
ag
e
an
d
i
m
p
r
o
v
e
a
Qu
alit
y
o
f
L
i
f
e
(
Qo
L
)
in
p
r
o
v
in
cial
ar
ea
s
.
No
te
t
h
at
t
h
ese
d
ata
w
as
p
ar
tiall
y
a
g
g
r
e
g
ated
b
y
te
m
p
o
r
ar
il
y
h
ir
e
d
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8708
I
n
t J
E
l
ec
&
C
o
m
p
E
n
g
,
Vo
l.
8
,
No
.
4
,
A
u
g
u
s
t 2
0
1
8
:
2
2
2
0
–
2
2
2
9
2224
s
taf
f
s
w
h
o
h
a
v
e
lo
w
er
s
k
ill
i
n
co
m
p
u
ter
li
ter
ac
y
.
Du
e
to
t
h
is
r
ea
s
o
n
it
led
to
n
o
is
y
a
n
d
m
is
s
i
n
g
d
ata
in
t
h
e
o
r
ig
in
al
d
ataset.
T
h
e
f
ir
s
t
d
ataset
co
n
tai
n
s
5
8
0
s
a
m
p
les
w
h
ile
t
h
e
s
ec
o
n
d
d
ataset
co
llects
1
9
8
s
a
m
p
les.
I
ts
attr
ib
u
tes
co
r
r
esp
o
n
d
to
th
e
p
u
b
lic
h
ea
lt
h
c
h
ar
ac
ter
is
tic
s
a
s
d
ep
icted
i
n
T
ab
le
1
,
s
o
m
e
p
ar
a
m
eter
s
s
u
ch
as
s
e
x
,
a
g
e,
a
n
d
o
th
er
s
w
h
ic
h
ar
e
n
eg
lecte
d
d
u
e
th
e
co
n
f
id
e
n
tialit
y
o
f
t
h
e
s
a
m
p
le’
s
o
w
n
er
,
h
o
w
ev
er
e
v
en
t
u
all
y
i
t
co
n
s
i
s
ts
o
f
in
v
e
s
ti
g
ati
v
e
attr
ib
u
te
s
w
h
ich
s
er
v
e
th
e
e
s
s
e
n
tial la
b
el
o
f
t
h
e
s
a
m
p
le.
T
ab
le
1
.
P
u
b
lic
Hea
lth
Data
s
e
t
A
ttrib
u
tes
A
t
t
r
i
b
u
t
e
Ty
p
e
R
o
l
e
G
e
n
d
e
r
N
o
mi
n
a
l
R
e
g
u
l
a
r
S
a
l
I
n
t
e
g
e
r
R
e
g
u
l
a
r
L
o
s
I
n
t
e
g
e
r
R
e
g
u
l
a
r
D
o
c
I
n
t
e
g
e
r
R
e
g
u
l
a
r
H
o
s
I
n
t
e
g
e
r
R
e
g
u
l
a
r
D
e
n
I
n
t
e
g
e
r
R
e
g
u
l
a
r
C
o
mm
e
n
t
s
N
o
mi
n
a
l
R
e
g
u
l
a
r
T
h
e
s
o
cc
er
p
ar
am
e
ter
s
ar
e
co
n
s
tr
u
cted
b
y
p
la
y
er
h
ei
g
h
t
(
H
E
I
)
,
w
ei
g
h
t
(
W
E
I
)
,
n
u
m
b
er
o
f
s
u
cc
ess
f
u
l
g
o
als
a
f
ter
th
o
u
s
a
n
d
atte
m
p
ts
(
GOL
)
,
n
u
m
b
er
o
f
p
ass
i
n
g
t
h
e
b
all
(
P
A
S)
an
d
a
v
er
ag
e
g
o
als
o
u
t
o
f
te
n
n
atio
n
al
co
m
p
eti
tio
n
s
(
A
VG)
.
So
cc
er
d
ataset
w
as
ca
p
tu
r
ed
b
y
a
m
ed
ia
co
m
p
a
n
y
w
h
ich
h
ad
b
ee
n
in
v
o
l
v
in
g
w
i
th
telev
i
s
io
n
b
r
o
ad
ca
s
tin
g
.
Da
ta
is
m
ea
n
t
to
h
elp
an
a
l
y
ze
p
la
y
er
’
s
p
er
f
o
r
m
an
ce
in
g
e
n
er
al.
No
te
th
at
th
e
s
e
d
ata
w
a
s
m
a
n
u
all
y
co
llected
b
y
c
o
m
p
a
n
y
s
ta
f
f
s
w
h
o
u
n
i
n
te
n
ti
o
n
all
y
d
ev
e
lo
p
ed
n
o
is
e
in
th
e
o
r
ig
in
al
d
ataset
a
s
w
ell.
I
t
s
attr
ib
u
te
s
ex
h
ib
it
t
h
e
p
lay
er
c
h
ar
ac
ter
is
tic
s
as
li
s
te
d
in
T
ab
le
2
,
s
o
m
e
p
ar
a
m
eter
s
s
u
c
h
a
s
ed
u
ca
tio
n
b
ac
k
g
r
o
u
n
d
,
in
co
m
e,
a
n
d
o
th
e
r
s
w
h
ich
ar
e
o
m
it
ted
d
u
e
t
h
e
s
ec
r
ec
y
o
f
th
e
p
la
y
er
’
s
i
n
f
o
r
m
a
tio
n
,
b
u
t
it c
o
n
tai
n
s
ex
p
er
i
m
e
n
tal
attr
ib
u
tes
w
h
ich
r
e
m
ar
k
t
h
e
s
i
g
n
if
ica
n
t la
b
e
l o
f
th
e
in
f
o
r
m
at
io
n
.
T
ab
le
2
.
So
cc
er
p
lay
er
Data
s
et
A
ttrib
u
tes
A
t
t
r
i
b
u
t
e
Ty
p
e
R
o
l
e
N
a
me
a
n
d
S
u
r
n
a
me
N
o
mi
n
a
l
R
e
g
u
l
a
r
T
e
a
m
N
o
mi
n
a
l
R
e
g
u
l
a
r
H
e
i
I
n
t
e
g
e
r
R
e
g
u
l
a
r
W
e
i
I
n
t
e
g
e
r
R
e
g
u
l
a
r
G
o
l
I
n
t
e
g
e
r
R
e
g
u
l
a
r
P
a
s
I
n
t
e
g
e
r
R
e
g
u
l
a
r
A
v
g
I
n
t
e
g
e
r
R
e
g
u
l
a
r
C
o
mm
e
n
t
s
N
o
mi
n
a
l
R
e
g
u
l
a
r
4
.
1
.
M
e
a
n a
bs
o
lute
er
ro
r
T
h
e
m
ea
n
ab
s
o
lu
te
er
r
o
r
(
MA
E
)
is
an
a
m
o
u
n
t
u
s
ed
to
q
u
an
ti
t
y
esti
m
ate
s
o
f
t
h
e
u
l
ti
m
a
te
r
es
u
lts
.
T
h
e
MA
E
i
s
a
m
ea
n
o
f
t
h
e
ab
s
o
lu
t
e
v
alu
e
o
f
f
la
w
s
an
d
ca
n
b
e
co
m
p
u
ted
b
y
:
∑
|
̂
|
(
1
)
w
h
er
e
is
t
h
e
d
ef
i
n
i
te
o
b
s
er
v
at
io
n
ti
m
e
s
er
ies a
n
d
̂
i
s
th
e
p
r
ed
icted
o
r
esti
m
ated
ti
m
e
s
er
ies.
4
.
2
.
Ro
o
t
m
ea
n sq
ua
re
d e
rr
o
r
R
o
o
t
m
ea
n
s
q
u
ar
ed
er
r
o
r
(
R
M
SE)
is
a
q
u
an
tit
y
u
s
ed
to
m
ea
s
u
r
e
th
e
d
if
f
er
en
ce
s
b
et
w
ee
n
s
a
m
p
le
an
d
p
o
p
u
latio
n
v
alu
e
s
f
o
r
ec
asted
b
y
a
m
o
d
el
o
r
est
i
m
ated
v
al
u
es
o
f
ac
t
u
al
o
b
s
er
v
at
io
n
s
.
T
h
e
R
MSE
d
en
o
tes
t
h
e
s
tan
d
ar
d
d
ev
iatio
n
o
f
t
h
e
d
if
f
er
en
ce
b
et
w
ee
n
f
o
r
ec
ast
s
a
n
d
o
b
s
er
v
atio
n
s
.
T
h
ese
d
if
f
er
en
c
es
ar
e
co
m
p
u
ted
b
y
th
e
s
a
m
p
le
d
ata
p
er
f
o
r
m
an
ce
o
v
er
p
r
ed
ictio
n
er
r
o
r
s
as c
alcu
lated
o
u
t
-
of
-
s
a
m
p
le.
T
h
e
R
MSE
o
f
f
o
r
ec
asted
v
alu
es
̂
f
o
r
ti
m
e
s
t
o
f
a
r
eg
r
es
s
io
n
's
d
ep
en
d
en
t v
ar
iab
le
is
ca
lc
u
lat
ed
f
o
r
n
d
if
f
er
en
t
f
o
r
ec
asts
a
s
s
h
o
w
n
in
E
q
u
atio
n
(
2
)
.
√
∑
̂
(
2
)
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:
2
0
8
8
-
8708
E
va
lu
a
tio
n
o
f a
Mu
ltip
le
R
e
g
r
ess
io
n
Mo
d
el
fo
r
N
o
is
y
a
n
d
Mis
s
in
g
Da
ta
(
C
h
a
n
in
to
r
n
Jitta
w
ir
iya
n
u
ko
o
n
)
2225
T
r
ain
in
g
a
d
ataset
i
n
g
en
er
al
w
il
l
d
ec
r
ea
s
e
th
e
er
r
o
r
r
ate
f
o
r
ex
p
er
i
m
en
t
s
et.
Fla
w
r
ate
f
o
r
tr
ain
in
g
d
ataset
is
r
elativ
el
y
h
i
g
h
er
th
an
th
at
o
f
t
h
e
ex
p
er
i
m
e
n
t
s
et.
I
f
an
y
t
w
o
alg
o
r
ith
m
s
r
es
u
lt
th
e
id
en
tical
M
A
E
th
en
R
MSE
i
s
tak
e
n
in
to
ac
co
u
n
t
f
o
r
s
elec
ti
n
g
t
h
e
f
in
e
s
t
alg
o
r
ith
m
.
No
r
m
al
l
y
e
x
p
er
i
m
e
n
ti
n
g
s
et
h
a
s
lo
w
f
la
w
ra
te
th
an
t
h
e
tr
ain
in
g
d
ataset.
4
.
3
.
No
is
e
a
nd
m
is
s
ing
pa
t
t
er
ns
Fu
ll
y
at
R
a
n
d
o
m
(
FaR
)
m
ea
n
s
n
o
i
s
e
o
r
m
i
s
s
i
n
g
p
atter
n
s
ar
e
n
o
t
d
ep
en
d
in
g
o
n
a
n
y
f
ac
to
r
s
.
Fo
r
in
s
ta
n
ce
m
a
n
y
q
u
esti
o
n
n
a
ir
es
w
ill
as
k
f
o
r
a
r
an
d
o
m
s
a
m
p
le.
I
n
ten
tio
n
a
ll
y
(
I
NT
)
m
ea
n
s
n
o
is
e
o
r
m
is
s
i
n
g
p
at
ter
n
s
r
ec
k
o
n
o
n
co
n
f
id
en
t
i
alities
.
Fo
r
ex
a
m
p
le,
s
o
m
e
r
esp
o
n
d
en
ts
m
a
y
a
w
k
w
ar
d
l
y
to
r
ep
o
r
t
th
eir
an
n
u
al
in
co
m
e,
a
g
e,
p
er
s
o
n
al
l
y
s
e
n
s
i
tiv
e
d
ata
etc.
E
v
e
n
tu
a
ll
y
th
e
y
m
a
y
f
ill
u
p
w
it
h
b
la
n
k
i
n
ten
tio
n
all
y
o
r
u
n
tr
u
e
f
i
g
u
r
es.
Mo
s
t
m
i
s
s
i
n
g
d
ata
p
atter
n
s
co
n
cl
u
d
e
ei
th
er
Fa
R
o
r
I
NT
.
4
.
4
.
M
a
dh
u t
re
a
t
m
ent
m
et
ho
d
Ma
d
h
u
a
n
d
Nag
ac
h
a
n
d
r
ik
a
[
1
6
]
p
r
esen
ted
tr
ea
t
m
en
t
m
et
h
o
d
f
o
r
d
ata
im
p
u
tatio
n
b
ased
u
p
o
n
d
u
al
d
is
tan
ce
v
ec
to
r
s
w
h
ich
ar
e
e
m
p
lo
y
ed
to
o
u
tlin
e
a
r
ep
r
esen
tat
io
n
b
et
w
ee
n
th
e
n
ea
r
es
t
n
ei
g
h
b
o
r
an
d
th
e
clu
s
ter
ce
n
tr
o
id
.
T
h
e
y
d
en
o
te
d
ataset
ele
m
en
ts
as
a
r
ep
r
esen
ta
tiv
e
p
x
q
m
a
tr
ix
.
Data
s
et
m
a
tr
ix
D
ch
ar
ac
ter
izes
t
h
e
ele
m
e
n
ts
o
f
p
r
o
w
s
an
d
q
co
l
u
m
n
s
an
d
ea
ch
r
o
w
co
n
tai
n
s
a
s
eq
u
en
ce
o
f
q
-
t
u
p
les
o
f
d
ata
elem
e
n
t
s
s
u
c
h
as
(
dk
1
,
d
k
2
,
d
k
3
,
…,
d
k
(
q
-
1)
,
d
kq
)
f
o
r
ea
ch
k
=
1
,
2
,
3
,
…,
p
.
A
ll
d
ataset
s
ar
e
ass
u
m
ed
t
o
b
e
a
s
et
o
f
f
in
i
te
ele
m
e
n
ts
.
An
ele
m
e
n
t
d
kq
is
c
o
n
s
id
er
ed
to
b
e
a
m
i
s
s
i
n
g
ele
m
en
t
w
h
en
e
v
er
{
d
ij
=
n
u
ll,
1
≤
i
≤
p
;
1
≤
j
≤
q
}
.
T
h
en
a
k
-
m
ea
n
s
alg
o
r
it
h
m
to
s
tr
u
ct
u
r
e
clu
s
ter
s
an
d
to
d
ef
in
e
th
e
ce
n
tr
o
id
s
u
s
in
g
t
h
e
v
ec
t
o
r
V
n
is
s
p
ec
if
ied
as
lis
ted
in
E
q
u
at
io
n
(
3
)
b
elo
w
:
|
|
∑
(
3
)
Hen
ce
,
t
h
e
n
ea
r
est
n
ei
g
h
b
o
r
b
ased
u
p
o
n
a
E
u
clid
ia
n
d
i
s
ta
n
ce
v
ec
to
r
w
ill
b
e
co
m
p
u
ted
f
o
r
m
is
s
in
g
v
alu
e
s
o
f
a
g
i
v
e
n
d
ataset.
A
s
s
u
m
e
t
h
at
D
is
a
s
et
o
f
f
i
n
ite
e
le
m
e
n
ts
an
d
b
o
th
m
a
n
d
n
co
r
r
esp
o
n
d
to
D
.
N
is
co
n
s
id
er
ed
to
b
e
th
e
n
ea
r
est
n
eig
h
b
o
r
o
f
M
i
f
an
d
o
n
l
y
i
f
N
is
th
e
n
ea
r
est
to
M
a
m
o
n
g
o
th
er
p
o
in
ts
lo
ca
ted
in
{D
–
M}
.
4
.
5
.
P
ro
po
s
ed
t
re
a
t
m
ent
m
et
ho
d
s
w
it
h less
bia
s
T
h
r
ee
p
r
o
p
o
s
ed
m
et
h
o
d
s
f
o
r
i
m
p
u
ti
n
g
d
ata
to
h
an
d
le
t
h
e
p
r
o
b
lem
o
f
n
o
is
y
an
d
m
i
s
s
i
n
g
v
alu
e
w
h
ic
h
ar
e
b
ased
u
p
o
n
lis
t
w
i
s
e
d
elet
io
n
,
ass
er
tio
n
,
an
d
r
an
d
o
m
iza
tio
n
h
a
v
e
b
ee
n
p
r
ese
n
ted
.
L
e
t
D
d
en
o
te
d
ataset
m
atr
i
x
w
h
ic
h
il
lu
s
tr
ates
a
r
ep
r
esen
tat
io
n
o
f
p
r
o
w
s
an
d
q
co
l
u
m
n
s
m
atr
i
x
(
dk
1
,
d
k
2
,
d
k
3
,
…
,
d
k
(
q
-
1)
,
d
kq
)
f
o
r
ea
ch
k
=
1
,
2
,
3
,
…,
p
.
T
h
e
d
ataset
is
a
s
s
u
m
ed
to
b
e
a
f
i
n
ite
s
et.
An
ele
m
e
n
t
d
kq
is
a
m
i
s
s
i
n
g
o
r
n
o
is
y
ele
m
e
n
t
(
NM
D)
w
h
e
n
e
v
er
{
d
ij
=
n
u
ll
|
|
n
o
is
e,
1
≤
i
≤
p
;
1
≤
j
≤
q
}
.
T
h
e
d
ataset
w
i
th
NM
D
ele
m
e
n
ts
is
ca
lled
u
n
e
x
ec
u
tab
le
d
ataset.
T
h
en
tr
ea
t
m
e
n
t
m
et
h
o
d
s
to
g
et
o
v
er
t
h
e
u
n
e
x
ec
u
tio
n
a
n
d
m
o
v
e
t
h
e
f
u
r
th
er
a
n
al
y
s
i
s
o
n
u
s
i
n
g
th
e
e
s
ti
m
ated
v
ec
to
r
E
n
ar
e
d
escr
ib
ed
in
th
e
f
o
llo
w
in
g
s
ec
tio
n
.
4
.
5
.
1
.
Dele
t
io
n
m
ec
ha
nis
m
(
DE
L
)
L
is
t
w
is
e
d
eletio
n
d
ea
ls
w
ith
t
h
e
NM
D
v
al
u
es
b
y
r
e
m
o
v
i
n
g
th
e
m
e
n
tire
l
y
in
o
r
d
er
th
at
d
ata
s
cien
ti
s
t
ca
n
an
al
y
ze
t
h
e
est
i
m
a
ted
d
ataset.
I
t
is
co
m
m
o
n
l
y
u
s
ed
m
e
th
o
d
an
d
r
ec
o
m
m
e
n
d
ed
w
h
e
n
th
e
m
is
s
in
g
n
es
s
is
Un
p
lan
n
ed
Mi
s
s
i
n
g
(
UM
)
ca
s
e.
DE
L
r
etain
s
t
h
e
h
u
m
b
le
an
d
s
i
m
p
le
tr
ea
t
m
e
n
t
tech
n
iq
u
e
w
h
et
h
er
o
r
n
o
t
th
e
NM
D
o
f
a
n
i
n
p
u
t
i
n
f
lu
e
n
ce
s
t
h
e
f
u
t
u
r
e
n
e
g
lecte
d
v
al
u
es.
An
y
z
r
o
w
s
o
f
m
atr
i
x
D
p
o
s
s
e
s
s
an
ele
m
en
t
d
ij
w
it
h
NM
D
w
h
er
e
{
d
ij
=
n
u
l
l
|
|
n
o
is
e,
1
≤
i
≤
p
;
1
≤
j
≤
q
}
th
en
th
e
e
n
tire
r
o
w
is
ca
n
ce
lled
.
T
h
e
esti
m
ated
E
n
d
ataset
is
{
d
ij
≠
n
u
ll
|
|
n
o
is
e
,
1
≤
i
≤
(
p
-
z
)
;
1
≤
j
≤
q
}
.
T
h
e
DE
L
tr
ea
t
m
e
n
t
i
s
k
n
o
wn
to
cu
lti
v
ate
a
f
air
p
r
ed
ictio
n
an
d
class
ical
a
n
al
y
tics
if
an
d
o
n
l
y
i
f
d
ata
s
et
i
s
lar
g
e,
w
h
er
e
p
o
w
e
r
i
s
tr
i
v
i
al
th
e
n
t
h
e
li
s
t
w
is
e
d
eletio
n
is
a
n
i
n
ter
es
tin
g
m
et
h
o
d
.
No
te,
th
e
s
t
u
d
y
i
n
t
h
i
s
p
ap
er
h
an
d
le
s
a
lar
g
e
s
a
m
p
le
a
n
d
th
e
a
s
s
u
m
p
t
io
n
o
f
UM
is
f
u
l
f
illed
t
h
en
t
h
i
s
d
eleti
o
n
is
d
elib
er
ated
to
b
e
an
ac
ce
p
tab
le
s
tr
ateg
y
.
4
.
5
.
2
.
Sin
g
le
a
s
s
er
t
io
n
m
ec
ha
nis
m
(
SAM
)
E
m
p
lo
y
d
u
m
m
y
v
ar
iab
le
n
a
m
el
y
a
v
er
ag
e
v
alu
e
to
i
m
p
u
te
d
ata
to
s
u
b
s
t
itu
te
th
e
m
is
s
in
g
n
e
s
s
.
Di
v
id
e
th
e
g
iv
e
n
D
d
ataset
i
n
to
2
g
r
o
u
p
s
th
a
t is:
a)
1
st
g
r
o
u
p
is
a
d
at
aset
w
h
ic
h
co
n
tain
s
ele
m
en
t
s
w
it
h
n
o
is
y
d
ata
(
N)
.
b
)
2
nd
g
r
o
u
p
r
ep
r
esen
ts
a
d
ataset
w
h
ich
co
n
tai
n
s
m
is
s
in
g
d
a
ta
(
MD
)
.
No
w
co
n
s
id
er
t
h
e
f
ir
s
t
g
r
o
u
p
as
g
ar
b
led
d
ataset
w
h
ic
h
i
s
u
n
e
x
ec
u
tab
le.
An
y
z
r
o
w
s
o
f
m
atr
i
x
D
p
o
s
s
es
s
an
e
le
m
e
n
t
d
ij
w
i
th
n
o
i
s
y
d
ata
(
N)
w
h
er
e
{
d
ij
=
n
o
is
e,
1
≤
i
≤
p
;
1
≤
j
≤
q
}
th
en
th
e
en
t
ir
e
r
o
w
is
ca
n
ce
lled
.
T
h
e
s
ec
o
n
d
g
r
o
u
p
d
at
aset
is
{
d
ij
≠
n
o
is
e,
1
≤
i
≤
(
p
-
z
)
;
1
≤
j
≤
q
}
.
T
h
e
s
u
b
s
t
itu
tio
n
f
o
r
esti
m
ated
En
d
ataset
w
it
h
d
ata
i
m
p
u
tatio
n
f
o
r
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8708
I
n
t J
E
l
ec
&
C
o
m
p
E
n
g
,
Vo
l.
8
,
No
.
4
,
A
u
g
u
s
t 2
0
1
8
:
2
2
2
0
–
2
2
2
9
2226
m
is
s
i
n
g
v
al
u
es i
s
d
ef
i
n
ed
as f
o
llo
w
s
:
|
|
∑
(
4
)
T
h
e
v
alid
atio
n
o
f
t
h
e
a
v
er
a
g
e
v
alu
e
is
t
h
at
it
is
an
ac
ce
p
tab
le
p
r
ed
ictio
n
f
o
r
a
r
a
n
d
o
m
p
ar
a
m
eter
o
u
t
o
f
a
n
o
r
m
al
d
is
tr
ib
u
t
io
n
.
I
n
ca
s
e
o
f
p
lan
n
ed
m
is
s
in
g
v
al
u
e,
t
h
is
tr
ea
t
m
e
n
t
m
e
th
o
d
w
i
ll
lea
d
an
u
n
p
r
ed
ictab
le
b
ias.
No
t o
n
l
y
t
h
is
m
et
h
o
d
d
ev
elo
p
s
d
is
ti
n
g
u
i
s
h
i
n
f
o
r
m
atio
n
b
u
t r
ath
er
g
r
o
w
s
t
h
e
s
ize
o
f
p
o
p
u
latio
n
co
m
p
ar
ed
to
DE
L
a
n
d
e
n
co
u
r
a
g
es
a
n
u
n
d
er
esti
m
ate
v
al
u
es.
Mo
r
eo
v
e
r
,
th
is
tech
n
iq
u
e
i
s
i
m
p
er
f
ec
t,
b
u
t
d
ev
e
lo
p
s
m
o
r
e
p
ar
am
eter
s
f
o
r
a
s
ca
le
s
co
r
e
r
ath
er
.
4
.
5
.
3
.
Ra
nd
o
m
m
et
ho
d (
RAM
)
Use
m
u
ltip
le
as
s
er
tio
n
s
,
m
ax
i
m
u
m
li
k
eli
h
o
o
d
at
r
an
d
o
m
f
o
r
r
ep
lace
m
en
t.
L
i
k
e
S
A
M,
th
e
D
d
ataset
m
u
s
t
b
e
s
p
lit
in
to
t
w
o
g
r
o
u
p
s
.
No
w
d
ea
l
w
it
h
th
e
1
st
g
r
o
u
p
as
co
r
r
u
p
ted
d
ataset
w
h
ich
ca
n
n
o
t
b
e
u
n
e
x
ec
u
tab
le.
An
y
z
r
o
w
s
o
f
m
atr
i
x
D
w
it
h
an
ele
m
en
t
o
f
d
ij
,
n
o
is
y
d
ata
(
N)
w
h
er
e
{
d
ij
=
n
o
i
s
e,
1
≤
i
≤
p
;
1
≤
j
≤
q
}
ar
e
t
h
en
r
e
m
o
v
ed
.
T
h
e
s
ec
o
n
d
g
r
o
u
p
d
ataset
is
{
d
ij
≠
n
o
is
e,
1
≤
i ≤
(
p
-
z
)
;
1
≤
j ≤
q
}
.
T
h
e
m
i
n
i
m
u
m
l
ik
eli
h
o
o
d
o
f
attr
ib
u
te
(
co
lu
m
n
)
j
(
w
h
er
e
j
=
1
,
2
,
3
,
…,
q
)
is
ch
ar
ac
ter
ized
b
y
d
(
m
i
n
)
j
w
h
er
e
d
(
m
i
n
)
j
=
Min
(
d
kj
)
f
o
r
ea
ch
k
=
1
,
2
,
3
,
…,
(
p
-
z
)
.
Si
m
i
lar
l
y
,
t
h
e
m
ax
i
m
u
m
li
k
el
ih
o
o
d
o
f
attr
ib
u
te
j
(
w
h
er
e
j
=
1
,
2
,
3
,
…,
q
)
is
r
ep
r
esen
ted
b
y
d
(
m
ax
)
j
w
h
er
e
d
(
m
ax
)
j
=
Ma
x
(
d
kj
)
f
o
r
ea
ch
k
=
1
,
2
,
3
,
…,
(
p
-
z
)
.
T
h
e
s
u
b
s
titu
tio
n
f
o
r
esti
m
ated
En
d
ataset
w
it
h
m
u
ltip
le
i
m
p
u
tatio
n
s
f
o
r
m
is
s
i
n
g
v
alu
e
s
in
ea
ch
attr
ib
u
te
j
is
r
an
d
o
m
l
y
d
eter
m
in
ed
as
f
o
llo
w
s
:
[
]
(
5
)
C
lear
l
y
t
h
e
p
r
o
p
o
s
ed
ap
p
r
o
ac
h
p
r
esen
t
s
co
lu
m
n
w
is
e
(
attr
ib
u
te
-
o
r
ien
tatio
n
)
o
p
er
atio
n
b
y
eli
m
i
n
ati
n
g
u
n
e
x
ec
u
tab
le
n
o
is
y
d
ata
t
h
en
i
m
p
u
ti
n
g
a
r
ep
lace
m
e
n
t
d
ata
b
ased
u
p
o
n
less
-
b
ias
-
m
ec
h
a
n
is
m
s
a
s
d
escr
ib
ed
ab
o
v
e.
Ho
w
e
v
er
,
Ma
d
h
u
’
s
m
eth
o
d
is
r
o
ww
is
e
(
i
n
s
tan
ce
o
r
ien
ted
)
o
p
er
atio
n
w
h
ic
h
ca
n
n
o
t
b
e
ap
p
lied
w
i
th
r
ea
lis
tic
ca
s
e
o
f
NM
D
f
o
r
t
w
o
r
ea
s
o
n
s
.
O
n
e
is
N
D
w
ill b
e
a
n
in
v
a
lid
f
i
g
u
r
e
in
s
tati
s
tical
ca
l
cu
latio
n
.
T
h
e
o
th
er
is
li
k
eli
h
o
o
d
in
ea
ch
attr
ib
u
te
is
m
o
r
e
s
ig
n
i
f
ica
n
t
t
h
a
n
i
n
s
t
an
ce
f
o
r
p
r
ed
ictin
g
a
f
u
tu
r
e
tr
en
d
.
A
co
m
p
ar
is
o
n
w
it
h
m
et
h
o
d
ex
p
lain
ed
in
[
1
6
]
is
d
is
p
la
y
ed
in
T
ab
le
3
.
T
ab
le
3
.
C
o
m
p
ar
is
o
n
w
i
th
E
x
i
s
tin
g
Me
t
h
o
d
D
A
TA
S
E
T
M
a
d
h
u
M
e
t
h
o
d
D
E
L
S
A
M
R
A
M
H
e
a
l
t
h
N
/
A
S
o
c
c
e
r
N
/
A
T
h
is
r
esear
ch
h
a
s
b
ee
n
co
n
d
u
cted
to
ca
r
r
y
o
u
t
a
n
i
n
-
d
ep
th
an
al
y
s
is
o
f
t
h
e
er
r
o
r
o
f
esti
m
atio
n
w
it
h
less
b
ias
(
DE
L
,
S
A
M
an
d
R
AM
)
co
m
p
ar
ab
le
to
t
w
o
o
r
ig
i
n
a
l
d
atasets
(
p
u
b
lic
h
ea
lt
h
an
d
s
o
cc
er
p
lay
er
s
)
.
T
h
e
s
tr
u
ct
u
r
es
o
f
t
h
ese
d
ataset
s
ar
e
as
lis
ted
in
T
ab
le
1
an
d
T
ab
le
2
.
T
a
b
le
4
d
ep
icts
th
e
o
v
er
all
r
esu
lts
f
o
r
co
r
r
elatio
n
co
ef
f
icie
n
t
(
C
OE
F),
m
ea
n
ab
s
o
l
u
te
er
r
o
r
(
M
A
E
)
an
d
r
o
o
t
m
ea
n
s
q
u
ar
ed
er
r
o
r
(
R
MSE
)
.
T
h
e
R
MSE
v
al
u
es
a
f
ter
ap
p
l
y
i
n
g
d
eletio
n
m
ec
h
a
n
is
m
(
DE
L
)
f
o
r
m
is
s
in
g
a
n
d
n
o
is
e
d
ata
co
m
p
ar
ab
l
y
d
if
f
er
a
n
d
th
e
y
ar
e
g
et
tin
g
lo
w
er
f
o
r
p
u
b
lic
h
ea
lt
h
d
ata
s
et.
W
h
e
n
co
m
p
ar
e
to
o
th
er
m
ec
h
an
i
s
m
s
,
t
h
e
MA
E
o
f
DE
L
ca
n
also
b
e
f
o
u
n
d
d
if
f
er
en
tl
y
,
th
i
s
is
th
e
lo
w
e
s
t
v
al
u
e
o
f
1
8
.
3
.
W
h
ile,
th
e
r
esu
lts
co
llected
f
o
r
w
h
e
n
th
e
s
o
cc
er
p
lay
er
d
ataset
h
as b
ee
n
tak
e
n
i
n
to
ac
co
u
n
t
f
o
r
th
e
e
v
alu
a
t
io
n
ar
e
f
air
l
y
clo
s
e.
T
ab
le
4
.
E
s
tim
a
t
io
n
w
it
h
R
o
o
t
Me
an
Sq
u
ar
e
E
r
r
o
r
D
A
TA
S
E
T
P
r
o
t
o
t
y
p
e
D
E
L
S
A
M
R
A
M
C
O
EF
M
A
E
R
M
S
E
C
O
EF
M
A
E
R
M
S
E
C
O
EF
M
A
E
R
M
S
E
C
O
EF
M
A
E
R
M
S
E
H
e
a
l
t
h
0
.
1
6
3
5
.
7
4
7
.
0
0
.
6
7
1
8
.
3
2
2
.
2
0
.
0
8
3
8
.
1
4
9
.
6
0
.
3
8
3
2
.
8
4
3
.
4
S
o
c
c
e
r
0
.
1
7
4
.
6
7
6
.
1
6
0
.
2
3
4
.
5
8
5
.
9
8
0
.
2
2
4
.
5
6
6
.
0
1
0
.
2
4
4
.
5
1
5
.
8
5
Dif
f
er
en
t
p
atter
n
s
o
f
u
n
r
ea
ll
y
s
h
ap
ed
m
i
s
s
i
n
g
o
b
s
er
v
atio
n
s
h
ad
b
ee
n
f
u
n
ctio
n
ed
f
o
r
t
w
o
m
en
tio
n
ed
d
atasets
.
On
ce
a
g
ai
n
,
t
h
e
h
u
m
b
le
d
eletio
n
i
s
in
ten
d
ed
to
p
r
o
b
e
th
e
s
en
s
it
iv
i
t
y
o
f
th
e
e
s
ti
m
atio
n
to
m
is
s
i
n
g
o
b
s
er
v
a
tio
n
s
i
n
ea
ch
attr
ib
u
t
e
an
d
th
e
ef
f
ec
t
o
f
d
eletio
n
o
n
co
r
r
elatio
n
co
ef
f
icie
n
t.
Fi
n
all
y
,
a
s
y
s
te
m
a
tic
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:
2
0
8
8
-
8708
E
va
lu
a
tio
n
o
f a
Mu
ltip
le
R
e
g
r
ess
io
n
Mo
d
el
fo
r
N
o
is
y
a
n
d
Mis
s
in
g
Da
ta
(
C
h
a
n
in
to
r
n
Jitta
w
ir
iya
n
u
ko
o
n
)
2227
p
atter
n
o
f
r
ep
lace
m
e
n
t
h
as b
ee
n
ex
p
lo
ited
to
co
m
p
en
s
ate
a
d
eletio
n
i
n
ca
s
e
o
f
m
i
s
s
i
n
g
d
ata
.
T
h
is
is
s
elec
ted
to
ec
h
o
th
e
ca
s
e
w
h
er
e,
f
o
r
in
s
ta
n
ce
,
co
n
f
id
en
tial
i
n
co
m
e
u
n
it
s
ar
e
m
o
r
e
h
esit
a
n
t
to
ex
p
lo
d
e
th
eir
i
n
co
m
e
in
t
h
e
r
ev
ie
w
.
I
n
m
o
s
t
ca
s
e
s
,
t
w
o
s
u
b
s
titu
t
io
n
p
atter
n
s
(
S
AM
a
n
d
R
A
M)
ar
e
e
x
er
cised
i
n
t
h
i
s
r
esear
ch
a
n
d
MO
A
s
i
m
u
lat
io
n
r
u
n
s
p
er
p
atter
n
s
a
r
e
ex
ec
u
ted
.
T
h
e
esti
m
ated
m
u
ltip
le
r
eg
r
es
s
io
n
eq
u
atio
n
s
with
t
h
eir
r
esp
ec
tiv
e
p
atter
n
s
f
o
r
n
o
i
s
e
an
d
m
is
s
in
g
d
ata
ar
e
s
i
m
u
la
ted
an
d
r
esu
lt
s
ar
e
s
u
m
m
ar
ized
i
n
T
ab
le
5
.
T
ab
le
5
.
Su
m
m
ar
y
o
f
M
u
ltip
le
R
eg
r
es
s
io
n
Mo
d
els
P
u
b
l
i
c
H
e
a
l
t
h
P
r
o
t
o
t
y
p
e
X
5
=
-
8
.
5
6
X
1
+
0
.
0
3
X
3
+
1
7
1
.
9
7
D
E
L
X
5
=
-
9
.
0
2
X
1
+
0
.
0
5
X
3
+
1
4
0
.
9
7
S
A
M
X
5
=
-
7
.
8
2
X
1
-
0
.
2
8
X
2
+
0
.
0
4
X
3
+
1
9
3
.
8
4
R
A
M
X
5
=
-
8
.
9
8
X
1
+
0
.
4
2
X
2
+
1
3
9
.
4
6
X
1
=
L
o
s,
X
2
=
D
o
c
,
X
3
=
H
o
s,
X
4
=
S
a
l
a
n
d
X
5
=
D
e
n
S
o
c
c
e
r
P
l
a
y
e
r
P
r
o
t
o
t
y
p
e
X
5
=
-
3
.
2
3
X
1
+
4
8
.
7
X
3
+
1
1
.
0
9
X
4
+
2
.
9
7
D
E
L
X
5
=
-
4
.
6
0
X
1
+
5
2
.
4
8
X
3
+
1
8
.
1
3
S
A
M
X
5
=
-
4
.
7
0
X
1
+
5
2
.
0
4
X
3
+
1
9
.
3
4
R
A
M
X
5
=
-
4
.
0
4
X
1
+
4
9
.
8
3
X
3
+
1
6
.
0
7
X
1
=
H
e
i
,
X
2
=
W
e
i
,
X
3
=
G
o
l
,
X
4
=
P
a
s
a
n
d
X
5
=
A
v
g
T
h
e
esti
m
ated
r
eg
r
ess
io
n
eq
u
atio
n
s
ca
n
h
e
n
ce
b
e
p
ar
alleled
w
it
h
th
e
r
ea
l/a
u
th
e
n
t
ic
d
ata
in
s
u
b
s
eq
u
en
t
y
ea
r
s
i
n
o
r
d
er
to
i
n
v
e
s
ti
g
ate
th
e
ac
cu
r
ac
y
o
f
ea
c
h
p
r
ed
ictio
n
.
As e
x
p
lai
n
ed
i
n
p
r
ev
io
u
s
s
ec
tio
n
,
t
h
e
m
u
ltip
le
r
eg
r
es
s
io
n
-
b
ased
i
m
p
u
tatio
n
f
o
r
m
s
a
f
o
r
ec
asti
n
g
tr
en
d
.
No
te
th
at
au
to
r
eg
r
ess
i
v
e
m
o
d
el
b
ased
i
m
p
u
ta
tio
n
ar
e
p
r
ese
n
ted
in
[
1
7
]
to
r
ec
k
o
n
m
is
s
i
n
g
v
al
u
es
an
d
th
e
ac
cu
r
ac
y
ar
e
ev
al
u
at
ed
b
y
u
s
in
g
R
MS
E
m
etr
ics
b
u
t
it
is
r
eg
ar
d
les
s
o
f
m
u
ltip
le
r
eg
r
es
s
io
n
m
o
d
el.
No
t
to
m
en
tio
n
th
e
y
d
o
n
o
t
co
n
ce
r
n
ab
o
u
t
th
e
ac
cu
r
ac
y
o
f
th
e
p
r
ed
ictio
n
tr
en
d
at
all.
W
h
ile
an
i
n
ter
es
tin
g
d
ata
i
m
p
u
ta
tio
n
f
o
r
esti
m
ati
n
g
t
h
e
m
is
s
in
g
v
al
u
e
is
in
tr
o
d
u
ce
d
in
[
1
6
]
,
h
o
w
e
v
er
th
is
n
e
w
p
ar
ad
ig
m
i
s
n
o
t
ap
p
licab
le
f
o
r
a
p
r
ed
ictio
n
m
o
d
el
as
w
ell.
T
h
at
is
,
it
ca
n
b
e
s
ee
n
as
th
e
f
ir
s
t
r
ea
s
o
n
t
h
at
th
e
i
m
p
u
ted
v
al
u
es
r
ef
l
ec
t
m
a
lf
u
n
ctio
n
.
T
h
i
s
p
r
o
b
lem
i
s
r
ep
r
esen
ted
i
n
Fig
u
r
e
3
an
d
Fi
g
u
r
e
4
.
T
h
e
a
u
th
en
tic
d
ata
p
o
in
t
s
d
i
v
er
g
e
f
r
o
m
t
h
e
r
e
g
r
ess
io
n
li
n
e
b
y
s
o
m
e
ex
te
n
t
b
u
t
i
m
p
u
ted
v
alu
e
s
j
a
m
p
er
f
ec
tl
y
o
n
th
e
r
e
g
r
ess
io
n
li
n
e.
T
h
is
p
r
o
b
lem
is
ea
s
il
y
f
i
x
ed
s
i
m
p
l
y
b
y
r
ep
laci
n
g
r
a
n
d
o
m
v
al
u
e
to
ea
ch
i
m
p
u
ted
v
al
u
e
(
th
is
r
ela
tes
to
ad
d
in
g
th
e
i
m
p
r
o
v
e
m
e
n
t)
.
T
h
e
s
ec
o
n
d
r
ea
s
o
n
th
er
e
is
s
li
g
h
t
v
ar
iab
ilit
y
s
h
ar
es
to
th
e
f
ac
t
t
h
at
t
h
e
r
e
g
r
ess
io
n
eq
u
atio
n
s
e
m
p
lo
y
ed
i
n
i
m
p
u
tat
io
n
ar
e
b
ased
u
p
o
n
a
s
a
m
p
le
f
r
o
m
to
tal
p
o
p
u
latio
n
.
A
s
ill
u
s
tr
ated
i
n
b
o
th
Fi
g
u
r
e
3
an
d
Fi
g
u
r
e
4
,
t
h
e
r
e
s
h
o
u
ld
b
e
ad
d
itio
n
al
co
m
p
e
n
s
atio
n
ar
o
u
n
d
t
h
e
d
o
tted
lin
ea
r
lin
e
in
o
r
d
er
to
lo
w
er
t
h
e
d
if
f
er
en
t
v
al
u
e
s
at
m
a
x
i
m
u
m
e
x
ten
t
.
Fig
u
r
e
3
.
C
o
m
p
ar
is
o
n
o
f
p
u
b
li
c
h
ea
lth
d
ata
w
it
h
au
th
e
n
tic
d
ata
Fig
u
r
e
4
.
C
o
m
p
ar
is
o
n
o
f
s
o
cc
er
p
lay
er
d
ata
w
ith
au
th
e
n
tic
d
ata
T
h
e
esti
m
atio
n
e
r
r
o
r
(
E
r
r
)
o
f
f
o
r
ec
asted
v
al
u
es
y
t
o
f
a
r
e
g
r
es
s
io
n
m
o
d
el
i
s
co
m
p
u
ted
b
y
co
m
p
ar
i
n
g
to
r
ea
l d
ata
x
t
as lis
ted
in
E
q
u
atio
n
(
6
)
.
|
|
(
6
)
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8708
I
n
t J
E
l
ec
&
C
o
m
p
E
n
g
,
Vo
l.
8
,
No
.
4
,
A
u
g
u
s
t 2
0
1
8
:
2
2
2
0
–
2
2
2
9
2228
Fig
u
r
e
5
p
r
esen
ts
p
er
ce
n
tag
e
er
r
o
r
o
f
p
r
e
d
ictio
n
u
s
i
n
g
m
u
ltip
le
r
eg
r
es
s
io
n
m
o
d
els
tab
u
lar
ized
in
T
ab
le
4
af
ter
co
m
p
ar
in
g
to
r
ea
l
d
ata
w
h
ic
h
h
as
o
cc
u
r
r
ed
s
u
b
s
eq
u
en
tl
y
.
Fi
g
u
r
e
5
r
ep
r
esen
ts
th
e
i
n
v
e
s
ti
g
atio
n
o
f
p
u
b
l
ic
h
ea
lt
h
d
ataset
w
h
ile
Fig
u
r
e
6
d
en
o
tes th
e
s
tu
d
y
o
f
s
o
cc
er
p
lay
er
d
ataset
r
esp
ec
tiv
e
l
y
Fig
u
r
e
5
.
E
s
ti
m
at
io
n
er
r
o
r
o
f
p
u
b
lic
h
ea
lt
h
d
ata
co
m
p
ar
i
n
g
to
a
u
th
e
n
tic
d
ata
Fig
u
r
e
6
.
E
s
ti
m
at
io
n
er
r
o
r
o
f
s
o
cc
er
p
lay
er
d
ata
co
m
p
ar
i
n
g
to
a
u
th
e
n
tic
d
a
ta
T
ab
le
6
s
u
m
m
ar
izes
t
h
e
av
e
r
ag
e
er
r
o
r
o
f
ea
ch
p
r
ed
ictio
n
b
y
ca
lc
u
lati
n
g
f
r
o
m
r
es
u
lt
s
s
h
o
w
n
in
Fig
u
r
e
5
an
d
Fig
u
r
e
6
.
I
t
is
a
p
p
ar
en
t
f
r
o
m
th
e
tab
le
DE
L
r
ef
lects
th
e
lo
w
est
av
er
a
g
e
p
er
ce
n
tag
e
o
f
er
r
o
r
an
d
attain
s
b
est ac
cu
r
ac
y
a
m
o
n
g
o
t
h
er
p
ar
ad
ig
m
s
.
T
ab
le
6
.
A
v
er
ag
e
E
r
r
o
r
f
o
r
I
n
d
iv
id
u
al
E
s
ti
m
atio
n
s
D
A
TA
S
E
T
A
v
e
r
a
g
e
Er
r
o
r
(
%)
P
r
o
t
o
t
y
p
e
D
E
L
S
A
M
R
A
M
H
e
a
l
t
h
3
3
.
5
6
2
8
.
9
7
3
5
.
1
5
3
4
.
7
7
S
o
c
c
e
r
4
2
.
7
4
4
0
.
7
4
2
.
7
3
4
4
.
0
1
5.
CO
NCLU
SI
O
N
T
h
e
r
esear
ch
ex
p
o
s
es
th
e
an
al
y
s
i
s
au
t
h
o
r
s
h
a
v
e
ad
o
p
ted
t
w
o
g
e
n
er
al
d
atasets
o
f
w
h
ic
h
b
o
t
h
ar
e
in
co
r
p
o
r
atin
g
m
i
s
s
i
n
g
d
ata
an
d
n
o
is
e.
T
o
g
et
o
v
er
th
e
li
m
it
atio
n
o
f
u
n
ex
c
u
tab
le
d
ataset
w
it
h
N
MD
,
N
a
n
d
MD
ar
e
d
iv
id
ed
i
n
to
t
w
o
p
ar
ts
.
Af
ter
r
e
m
o
v
i
n
g
in
v
alid
f
ig
u
r
e,
t
h
e
tr
ea
t
m
en
t
m
eth
o
d
s
w
h
ic
h
c
u
s
to
m
iz
e
esti
m
atio
n
w
it
h
les
s
b
ias
e
x
er
cisi
n
g
d
eletio
n
,
s
i
n
g
le
as
s
er
ti
o
n
an
d
r
an
d
o
m
m
ec
h
a
n
is
m
s
a
r
e
p
r
o
p
o
s
ed
.
T
h
ese
ar
e
o
f
s
i
m
p
li
f
ied
to
o
ls
a
n
d
u
n
s
u
b
s
tan
tial
b
ias
t
y
p
e
s
.
T
h
e
k
e
y
p
o
in
t
o
f
c
u
r
r
en
t
in
v
e
s
tig
a
tio
n
is
t
h
at
a
p
r
o
to
ty
p
e’
s
f
i
le
av
ail
s
i
n
o
r
d
er
to
co
p
e
w
it
h
t
h
e
m
is
s
i
n
g
n
es
s
.
P
r
o
p
o
s
ed
m
ec
h
an
i
s
m
s
ar
e
co
n
s
ti
tu
ted
to
r
ef
o
r
m
th
e
m
is
s
in
g
d
ata
a
n
d
th
e
v
er
if
icatio
n
is
p
er
f
o
r
m
ed
o
n
t
h
i
s
ad
j
u
s
t
m
e
n
t
al
s
o
.
T
h
e
p
er
f
o
r
m
an
ce
o
f
d
eletio
n
m
ec
h
a
n
i
s
m
i
s
g
o
o
d
o
n
b
o
th
d
at
asets
as
co
m
p
ar
ed
to
th
e
r
est.
R
e
m
ar
k
ab
le
p
o
in
t
is
th
a
t
MA
E
an
d
R
M
SE
r
e
m
ain
al
m
o
s
t
id
en
t
icall
y
i
n
a
ll
th
e
t
w
o
d
ataset
s
.
T
h
e
u
n
iq
u
e
asp
ec
ts
o
f
th
e
co
n
s
ta
n
c
y
o
f
m
is
s
i
n
g
p
atter
n
s
ar
e
p
er
ce
iv
ed
in
th
e
c
u
r
r
en
t
s
t
u
d
y
.
T
h
is
is
th
e
lead
in
g
ti
m
e
i
n
th
e
liter
atu
r
e
s
i
n
ce
n
o
g
e
n
er
al
d
ataset
in
p
r
ac
tice
i
s
o
b
tain
ab
le
in
th
e
f
o
r
m
o
f
m
is
s
in
g
t
y
p
e
s
r
ath
er
th
a
n
th
e
r
e
-
co
n
s
tr
u
ct
io
n
o
f
d
ataset
s
in
t
h
e
MO
A
s
i
m
u
latio
n
m
o
d
el.
Fro
m
th
e
s
e
r
esu
lts
it i
s
f
o
u
n
d
th
at,
t
h
e
ess
e
n
tia
l f
i
g
u
r
es o
f
av
er
a
g
e
er
r
o
r
ar
e
ap
p
r
o
p
r
iate
to
j
u
d
g
e
w
h
ic
h
m
ec
h
a
n
i
s
m
s
ca
n
b
e
o
p
ted
f
o
r
th
e
s
ak
e
o
f
f
u
ll
r
ea
ch
.
Fro
m
t
ab
le
ab
o
v
e,
w
e
ca
n
o
b
v
io
u
s
l
y
s
tate
th
at
DE
L
h
as
ac
h
iev
ed
t
h
e
b
est
ac
cu
r
ac
y
a
m
o
n
g
o
th
er
s.
T
h
e
f
o
r
th
co
m
in
g
w
o
r
k
w
ill
co
n
s
id
er
th
e
ad
d
itio
n
o
f
th
e
p
r
esen
t
r
esear
ch
f
o
c
u
s
i
n
g
o
n
d
i
f
f
er
e
n
t
v
ar
iatio
n
s
o
f
n
o
is
e
d
ata
s
ets
ap
p
r
o
ac
h
in
MO
A
.
T
h
e
n
e
x
t
p
ap
er
w
ill
g
r
ip
a
co
n
ce
p
t o
f
d
ep
r
ess
in
g
an
er
r
o
r
p
er
ce
n
tag
e
o
f
r
eg
r
es
s
i
v
e
esti
m
atio
n
i
n
MO
A
.
RE
F
E
R
E
NC
E
S
[1
]
K.
S
u
,
e
t
a
l
.
,
“
S
m
a
rt
Cit
y
a
n
d
t
h
e
A
p
p
li
c
a
ti
o
n
s,”
I
n
ter
n
a
ti
o
n
a
l
Co
n
fer
e
n
c
e
o
n
El
e
c
tro
n
ics
,
Co
m
mu
n
ica
t
io
n
s
a
n
d
Co
n
tro
l
(
ICECC)
,
p
p
.
1
0
2
8
-
1
0
3
1
,
2
0
1
1
.
[2
]
A
.
P
ra
b
a
h
a
r,
“
De
v
e
lo
p
m
e
n
t
o
f
Hig
h
P
e
rf
o
rm
a
n
c
e
W
irele
s
s
S
e
n
so
r
No
d
e
f
o
r
A
c
o
u
stic
A
p
p
li
c
a
ti
o
n
s,”
IEE
E
In
ter
n
a
t
io
n
a
l
C
o
n
fer
e
n
c
e
o
n
Gr
e
e
n
Hig
h
Per
fo
rm
a
n
c
e
Co
mp
u
ti
n
g
(
ICGH
PC)
,
p
p
.
1
-
5
,
2
0
1
3
.
[3
]
V
.
M
a
rx
,
“
T
h
e
Big
Ch
a
ll
e
n
g
e
o
f
Big
Da
ta,”
Na
tu
re
4
9
8
.
7
4
5
3
,
p
p
.
2
5
5
-
2
6
0
,
2
0
1
3
.
[4
]
C
.
En
d
e
rs,
“
A
p
p
li
e
d
M
issi
n
g
Da
ta A
n
a
l
y
sis,”
G
u
il
f
o
rd
P
re
ss
,
Ne
w
Yo
rk
,
2
0
1
0
.
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:
2
0
8
8
-
8708
E
va
lu
a
tio
n
o
f a
Mu
ltip
le
R
e
g
r
ess
io
n
Mo
d
el
fo
r
N
o
is
y
a
n
d
Mis
s
in
g
Da
ta
(
C
h
a
n
in
to
r
n
Jitta
w
ir
iya
n
u
ko
o
n
)
2229
[5
]
H.
El
d
ird
iery
a
n
d
A
.
H.
A
h
m
e
d
,
“
De
tec
ti
n
g
a
n
d
Re
m
o
v
in
g
No
is
y
Da
ta
o
n
Web
Do
c
u
m
e
n
t
u
sin
g
Tex
t
D
e
n
sity
A
p
p
ro
a
c
h
,
”
In
ter
n
a
ti
o
n
a
l
J
o
u
rn
a
l
o
f
C
o
mp
u
ter
Ap
p
li
c
a
ti
o
n
s
,
v
ol
.
1
1
2
,
n
o
.
5
,
p
p
.
3
2
-
2
6
,
2
0
1
5
.
[6
]
R.
S
h
a
rm
a
a
n
d
M
.
B
h
a
ti
a
,
“
E
li
m
in
a
ti
n
g
th
e
No
ise
f
ro
m
We
b
P
a
g
e
s
u
sin
g
P
a
g
e
Re
p
lac
e
m
e
n
t
A
lg
o
rit
h
m
,
”
In
ter
n
a
t
io
n
a
l
J
o
u
r
n
a
l
o
f
Co
m
p
u
t
e
r
S
c
ien
c
e
a
n
d
I
n
fo
rm
a
ti
o
n
T
e
c
h
n
o
l
o
g
ies
(
IJ
CS
IT
)
,
v
ol
.
5
,
n
o
.
3
,
p
p
.
3
0
6
6
-
3068
,
2
0
1
4
.
[7
]
B.
Ch
a
e
,
e
t
a
l
.
,
“
T
h
e
im
p
a
c
t
o
f
a
d
v
a
n
c
e
d
a
n
a
ly
ti
c
s
a
n
d
d
a
ta
a
c
c
u
ra
c
y
o
n
o
p
e
ra
ti
o
n
a
l
p
e
rf
o
rm
a
n
c
e
:
A
c
o
n
ti
n
g
e
n
t
re
so
u
rc
e
b
a
se
d
th
e
o
ry
(RB
T
)
p
e
rsp
e
c
ti
v
e
,
”
De
c
isio
n
S
u
p
p
o
rt S
y
ste
ms
,
v
o
l.
5
9
,
p
p
.
1
1
9
-
1
2
6
,
2
0
1
4
.
[8
]
M
.
A
.
S
h
a
b
ir
a
n
d
P
.
T
.
De
e
p
a
li
,
“
S
a
telli
te
Im
a
g
e
De
n
o
isi
n
g
Us
in
g
Disc
re
te
Co
sin
e
T
ra
n
s
f
o
rm
,
”
In
d
o
n
e
sia
n
J
o
u
rn
a
l
o
f
El
e
c
trica
l
En
g
in
e
e
rin
g
a
n
d
In
f
o
rm
a
ti
c
s
,
v
o
l.
5
,
p
p
.
3
72
-
3
7
5
,
2
0
1
7
.
[9
]
A
.
Dh
a
r
m
a
ra
j
a
n
a
n
d
T
.
V
e
lm
u
ru
g
a
n
,
“
L
u
n
g
Ca
n
c
e
r
Da
ta
A
n
a
ly
sis
b
y
k
-
m
e
a
n
s
a
n
d
F
a
rth
e
stF
irst
Clu
ste
rin
g
A
l
g
o
rit
h
m
s
,
”
In
d
ia
n
J
o
u
rn
a
l
o
f
S
c
ien
c
e
a
n
d
T
e
c
h
n
o
lo
g
y
,
v
ol
.
8
,
n
o
.
15
,
2
0
1
5
.
[1
0
]
A
.
Bi
fe
t,
e
t
a
l
.
,
“
M
O
A
:
M
a
s
siv
e
On
li
n
e
A
n
a
l
y
sis
,
”
J
o
u
rn
a
l
o
f
M
a
c
h
in
e
L
e
a
rn
i
n
g
Res
e
a
rc
h
,
v
o
l.
1
1
,
p
p
.
1
6
0
1
-
1
6
0
4
,
2
0
1
0
.
[1
1
]
Z.
B
.
Yo
ss
e
f
a
n
d
a
n
d
S
.
Ra
jag
o
p
a
lan
,
“
T
e
m
p
late
De
te
c
ti
o
n
v
ia
Da
ta
M
in
in
g
a
n
d
Its
A
p
p
li
c
a
ti
o
n
s,”
Pro
c
e
e
d
in
g
s
o
f
th
e
In
ter
n
a
ti
o
n
a
l
C
o
n
fer
e
n
c
e
o
n
t
h
e
W
o
rld
W
id
e
W
e
b
,
A
CM
P
re
ss
,
p
p
.
5
8
0
-
5
9
1
,
2
0
0
2
.
[1
2
]
L
.
Yi,
e
t
a
l
.
,
“
El
im
in
a
ti
n
g
No
isy
In
f
o
rm
a
ti
o
n
i
n
W
e
b
P
a
g
e
s
f
o
r
Da
ta
M
in
i
n
g
,
”
S
IGKD
D
,
A
CM
P
re
ss
,
p
p
.
2
9
6
-
3
0
5
,
2
0
0
3
.
[1
3
]
S
.
De
b
n
a
th
,
e
t
a
l
.
,
“
A
u
to
m
a
ti
c
E
x
trac
t
io
n
o
f
In
f
o
r
m
a
ti
v
e
Blo
c
k
s
f
ro
m
W
e
b
p
a
g
e
s,”
ACM
S
y
mp
o
siu
m
o
n
A
p
p
li
e
d
Co
mp
u
t
in
g
,
p
p
.
1
7
2
2
-
1
7
2
6
,
2
0
0
5
.
[1
4
]
E.
S
.
L
a
b
e
r,
e
t
a
l
.
,
“
F
a
st
a
n
d
S
imp
le
M
e
th
o
d
f
o
r
Ex
trac
ti
n
g
Re
le
v
a
n
t
Co
n
ten
t
f
ro
m
N
e
w
s
W
e
b
p
a
g
e
s
,
”
Pro
c
e
e
d
in
g
s
o
f
t
h
e
ACM
Co
n
fer
e
n
c
e
o
n
I
n
fo
r
ma
ti
o
n
a
n
d
Kn
o
wl
e
d
g
e
M
a
n
a
g
e
me
n
t
,
A
CM
P
re
ss
,
p
p
.
1
6
8
5
-
1
6
8
8
,
2
0
0
9
.
[1
5
]
N
.
Ra
h
e
ja
a
n
d
V
.
K.
Ka
ti
y
a
r,
“
No
ise
Re
d
u
c
ti
o
n
A
p
p
ro
a
c
h
Ba
se
d
o
n
n
x
1
T
a
b
le
a
n
d
X
S
L
Disp
la
y
M
e
th
o
d
f
o
r
Eff
icie
n
t
W
e
b
Da
ta E
x
tra
c
ti
o
n
,
”
I
n
ter
n
a
ti
o
n
a
l
J
o
u
rn
a
l
o
f
C
o
mp
u
ter
Ap
p
li
c
a
ti
o
n
s
,
v
o
l.
6
4
,
2
0
1
3
.
[1
6
]
G
.
M
a
d
h
u
a
n
d
G
.
Na
g
a
c
h
a
n
d
rik
a
,
“
A
Ne
w
P
a
ra
d
ig
m
f
o
r
De
v
e
lo
p
m
e
n
t
o
f
D
a
ta
I
m
p
u
tatio
n
A
p
p
ro
a
c
h
f
o
r
M
issin
g
V
a
lu
e
Esti
m
a
ti
o
n
,
”
I
n
ter
n
a
ti
o
n
a
l
J
o
u
rn
a
l
o
f
El
e
c
trica
l
a
n
d
C
o
mp
u
t
e
r E
n
g
i
n
e
e
rin
g
,
v
o
l.
6
,
p
p
.
3
2
2
2
-
3
2
2
8
,
2
0
1
6
.
[1
7
]
R.
T
h
iru
m
a
h
a
l
a
n
d
P
.
A
.
De
e
p
a
l
i,
“
KN
N
a
n
d
A
R
L
Ba
se
d
I
m
p
u
tat
io
n
t
o
Esti
m
a
te
M
issin
g
V
a
lu
e
s,”
In
d
o
n
e
sia
n
J
o
u
rn
a
l
o
f
El
e
c
trica
l
En
g
in
e
e
rin
g
a
n
d
I
n
fo
rm
a
t
ics
,
v
o
l.
2
,
p
p
.
1
1
9
-
1
2
4
,
2
0
1
4
.
Evaluation Warning : The document was created with Spire.PDF for Python.