I
nte
rna
t
io
na
l J
o
ur
na
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.
10
,
No
.
1
,
Feb
r
u
ar
y
2020
,
p
p
.
415
~
42
0
I
SS
N:
2
0
8
8
-
8708
,
DOI
: 1
0
.
1
1
5
9
1
/
i
j
ec
e
.
v
1
0
i
1
.
pp
4
1
5
-
420
415
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ij
ec
e.
ia
esco
r
e.
co
m/in
d
ex
.
p
h
p
/I
JE
C
E
A
t
i
m
e
e
f
ficien
t
a
nd
a
ccurate
r
et
rie
v
a
l of
r
a
ng
e
a
g
g
rega
te
q
ueries using
f
u
zzy
c
lustering
m
ea
n
s (F
CM
)
a
ppro
a
c
h
A.
M
urug
a
n,
D.
G
o
bin
a
t
h,
S.
G
a
nes
h K
u
m
a
r,
B.
M
urug
a
na
ntha
m
,
Sa
ruba
la
Velus
a
my
De
p
a
rtme
n
t
o
f
Co
m
p
u
ter S
c
ien
c
e
a
n
d
E
n
g
in
e
e
rin
g
,
F
a
c
u
l
ty
o
f
En
g
in
e
e
rin
g
a
n
d
T
e
c
h
n
o
lo
g
y
,
S
RM
IS
T
,
Ch
e
n
n
a
i
,
I
n
d
ia
Art
icle
I
nfo
AB
ST
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
Feb
17
,
2
0
1
9
R
ev
i
s
ed
A
u
g
1
0
,
20
19
A
cc
ep
ted
A
u
g
29
,
2
0
19
M
a
ss
iv
e
g
ro
w
th
in
th
e
b
ig
d
a
ta
m
a
k
e
s
di
ff
icu
lt
to
a
n
a
l
y
s
e
a
n
d
re
tri
e
v
e
th
e
u
se
f
u
l
in
f
o
r
m
a
ti
o
n
f
ro
m
th
e
se
t
o
f
a
v
a
il
a
b
le
d
a
ta’s
.
Ex
isti
n
g
a
p
p
ro
a
c
h
e
s
c
a
n
n
o
t
g
u
a
ra
n
tee
a
n
e
ff
icie
n
t
re
tri
e
v
a
l
o
f
d
a
ta
f
ro
m
th
e
d
a
tab
a
se
.
In
th
e
e
x
isti
n
g
w
o
rk
stra
ti
f
ied
sa
m
p
li
n
g
is
u
se
d
to
p
a
rti
ti
o
n
th
e
tab
les
in
ter
m
s
o
f
stra
ti
c
v
a
riab
les
.
Ho
w
e
v
e
r
k
m
e
a
n
s
c
lu
ste
rin
g
a
lg
o
rit
h
m
c
a
n
n
o
t
g
u
a
ra
n
tee
s
a
n
e
ff
icie
n
t
re
tri
e
v
a
l
w
h
e
re
th
e
c
h
o
o
si
n
g
c
e
n
tro
id
in
t
h
e
larg
e
v
o
lu
m
e
o
f
d
a
ta
w
o
u
ld
b
e
d
if
f
icu
lt
.
A
n
d
les
s
k
n
o
w
led
g
e
a
b
o
u
t
t
h
e
stra
ti
c
v
a
riab
le
m
i
g
h
t
lea
d
s
to
th
e
les
s
e
f
f
i
c
i
e
n
t
p
a
rti
ti
o
n
i
n
g
o
f
tab
les
.
T
h
is
p
ro
b
lem
is
o
v
e
rc
o
m
e
in
th
e
p
ro
p
o
se
d
m
e
th
o
d
o
lo
g
y
b
y
in
tro
d
u
c
in
g
th
e
F
CM
c
lu
ste
rin
g
i
n
ste
a
d
o
f
k
m
e
a
n
s
c
lu
ste
rin
g
w
h
ich
c
a
n
c
lu
ste
r
th
e
larg
e
v
o
lu
m
e
o
f
d
a
ta
w
h
ich
a
re
sim
il
a
r
in
n
a
tu
re
.
S
tratif
ica
ti
o
n
p
ro
b
lem
is
o
v
e
rc
o
m
e
b
y
in
tro
d
u
c
in
g
th
e
p
o
st
stra
ti
f
ica
ti
o
n
a
p
p
ro
a
c
h
w
h
ich
w
il
l
lea
d
s
to
e
ff
icie
n
t
se
lec
ti
o
n
o
f
stra
ti
c
v
a
riab
le.
T
h
is
m
e
th
o
d
o
l
o
g
y
lea
d
s
to
a
n
e
ff
icie
n
t
re
tri
e
v
a
l
p
ro
c
e
ss
in
term
s
o
f
u
se
r
q
u
e
ry
w
it
h
in
les
s ti
m
e
a
n
d
m
o
re
a
c
c
u
ra
c
y
.
K
ey
w
o
r
d
s
:
B
ig
d
ata
FC
M
al
g
o
r
ith
m
P
o
s
t stra
tif
icatio
n
s
a
m
p
lin
g
Co
p
y
rig
h
t
©
2
0
2
0
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
:
A.
M
u
r
u
g
an
,
Dep
ar
t
m
en
t o
f
C
o
m
p
u
ter
Scie
n
ce
an
d
E
n
g
i
n
ee
r
in
g
,
Facu
lt
y
o
f
E
n
g
i
n
ee
r
i
n
g
a
n
d
T
e
ch
n
o
lo
g
y
,
SR
MI
ST
,
C
h
e
n
n
ai
,
I
n
d
ia.
E
m
ail:
a
m
u
r
u
g
an
1
5
@
o
u
tlo
o
k
.
co
m
1.
I
NT
RO
D
UCT
I
O
N
B
ig
d
ata
an
al
y
s
i
s
ca
n
r
ev
ea
l
t
h
at
t
h
e
co
s
t
to
p
r
o
ce
s
s
t
h
e
d
ata
in
b
i
g
d
ata
en
v
ir
o
n
m
en
t
is
to
o
h
ig
h
,
s
u
c
h
as
p
ee
r
in
f
l
u
en
ce
a
m
o
n
g
cu
s
to
m
er
s
,
d
is
clo
s
ed
b
y
a
n
al
y
s
i
n
g
s
h
o
p
p
er
’
s
tr
an
s
ac
tio
n
s
,
s
o
cial
an
d
g
eo
g
r
ap
h
ical
d
ata.
I
t
i
s
a
ch
a
llen
g
in
g
tas
k
i
n
b
i
g
d
ata
en
v
i
r
o
n
m
e
n
t
to
q
u
ic
k
l
y
ac
q
u
ir
e
t
h
e
r
es
u
lt
f
o
r
r
a
n
g
e
-
ag
g
r
e
g
ate
q
u
er
ie
s
.
A
r
an
g
e
q
u
er
y
i
s
a
co
m
m
o
n
d
atab
ase
o
p
er
atio
n
th
at
r
etr
ie
v
es
all
r
ec
o
r
d
s
f
r
o
m
d
atab
ase
w
h
er
e
s
o
m
e
v
alu
e
e
x
is
ts
b
et
wee
n
an
u
p
p
er
an
d
lo
w
er
b
o
u
n
d
ar
y
.
Fo
r
ex
a
m
p
le,
li
s
t
all
e
m
p
lo
y
ee
s
w
ith
3
to
5
y
ea
r
s
o
f
e
x
p
er
ien
ce
.
R
an
g
e
q
u
er
ies
ar
e
u
n
u
s
u
al
b
ec
au
s
e
w
e
d
o
n
’
t
k
n
o
w
i
n
ad
v
an
ce
th
a
t
h
o
w
m
a
n
y
r
o
w
s
w
ill
b
e
r
etu
r
n
ed
a
s
r
es
u
lt
.
I
t
w
ill
r
etu
r
n
an
y
n
u
m
b
er
o
f
r
o
w
s
o
r
n
o
t
ev
e
n
a
n
y
.
Ma
n
y
o
t
h
er
q
u
er
i
es,
s
u
ch
as
th
e
to
p
ten
m
o
s
t
s
e
n
io
r
e
m
p
lo
y
ee
s
,
o
r
th
e
n
e
w
e
s
t
e
m
p
lo
y
ee
,
ca
n
b
e
d
o
n
e
m
o
r
e
ef
f
icie
n
tl
y
b
ec
a
u
s
e
th
ese
q
u
er
ie
s
h
a
v
e
an
u
p
p
er
b
o
u
n
d
to
t
h
e
n
u
m
b
er
o
f
r
esu
l
ts
t
h
e
y
w
ill
r
et
u
r
n
.
R
a
n
g
e
q
u
er
y
r
etr
ie
v
al
i
n
th
e
s
m
al
ler
d
atab
ase
w
o
u
ld
b
e
m
o
r
e
ef
f
icie
n
t
p
r
o
ce
s
s
w
h
er
e
it
w
ill
co
n
s
u
m
e
m
o
r
e
ti
m
e
an
d
w
ill
r
ed
u
ce
th
e
ti
m
e
co
m
p
lex
i
t
y
i
n
ca
s
e
o
f
lar
g
er
d
atab
ase
w
h
er
e
t
h
e
b
i
g
v
o
l
u
m
e
o
f
r
ec
o
r
d
s
ar
e
p
r
esen
t.
T
h
is
p
r
o
b
le
m
i
s
ad
d
r
ess
es
i
n
t
h
e
ex
is
ti
n
g
m
et
h
o
d
o
lo
g
y
in
ter
m
s
o
f
r
etr
iev
i
n
g
m
o
r
e
ac
cu
r
ate
a
n
d
tim
e
co
n
ce
r
n
ed
o
u
tp
u
t
f
o
r
th
e
r
an
g
e
q
u
er
ies
f
r
o
m
th
e
lar
g
er
d
atab
ases
.
I
n
b
ig
d
ata
en
v
ir
o
n
m
e
n
t
m
o
s
tl
y
th
e
d
ata
s
ets
ar
e
s
to
r
ed
in
d
i
s
tr
ib
u
ted
en
v
ir
o
n
m
e
n
t
as
t
h
e
v
o
lu
m
e
o
f
th
e
d
ata
i
s
v
er
y
lar
g
e.
T
h
e
co
s
t
o
f
d
is
tr
ib
u
ted
r
an
g
e
-
ag
g
r
eg
ate
q
u
er
ies
m
a
in
l
y
d
ep
en
d
s
o
n
t
w
o
f
ac
to
r
s
.
i.e
.
,
n
et
w
o
r
k
co
m
m
u
n
icatio
n
co
s
t
an
d
lo
ca
l
f
iles
s
ca
n
n
i
n
g
co
s
t.
T
h
e
f
ir
s
t
co
s
t
d
ep
en
d
s
o
n
d
ata
tr
an
s
m
is
s
io
n
an
d
s
y
n
c
h
r
o
n
iza
tio
n
f
o
r
a
g
g
r
eg
ate
o
p
er
atio
n
s
i
f
t
h
e
f
ile
s
s
elec
ted
ar
e
in
d
if
f
er
en
t
s
er
v
er
s
.
T
h
e
s
ec
o
n
d
co
s
t
d
ep
en
d
s
o
n
lo
ca
l
f
ile
s
s
ca
n
n
in
g
to
s
ea
r
ch
t
h
e
s
e
lecte
d
tu
p
les.
W
h
en
th
e
d
ata
s
et
s
ize
g
et
s
i
n
cr
ea
s
ed
co
n
tin
u
o
u
s
l
y
,
t
h
e
ab
o
v
e
m
e
n
ti
o
n
ed
t
w
o
co
s
ts
w
il
l
also
g
et
i
n
cr
ea
s
e
d
r
am
atica
ll
y
.
So
o
n
l
y
w
h
e
n
th
e
ab
o
v
e
t
w
o
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8708
I
n
t J
E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
10
,
No
.
1
,
Feb
r
u
ar
y
2
0
2
0
:
415
-
420
416
co
s
ts
ar
e
m
i
n
i
m
ized
,
w
e
ca
n
g
et
th
e
f
in
al
r
an
g
e
-
a
g
g
r
eg
at
e
q
u
er
ies
r
es
u
lt
f
aster
i
n
b
i
g
d
ata
en
v
ir
o
n
m
e
n
ts
.
T
h
er
e
ar
e
v
ar
io
u
s
ap
p
r
o
ac
h
e
s
to
m
i
n
i
m
ize
t
h
e
ab
o
v
e
t
wo
co
s
ts
t
h
at
h
as
b
ee
n
s
t
u
d
ied
in
t
h
e
p
ast
[
1
-
4
]
,
b
u
t
to
d
ay
a
n
al
y
s
t
s
ar
e
p
r
o
g
r
e
s
s
i
v
el
y
m
o
r
e
f
o
cu
s
in
g
o
n
e
f
f
icie
n
t
m
eth
o
d
s
a
n
d
to
o
ls
f
o
r
b
ig
d
ata
an
al
y
s
i
s
.
T
h
e
s
a
m
p
li
n
g
a
n
d
h
is
to
g
r
am
ap
p
r
o
ac
h
es
w
er
e
alr
ea
d
y
u
s
ed
in
b
id
d
ata
e
n
v
ir
o
n
m
en
ts
to
s
u
p
p
o
r
t
ap
p
r
o
x
im
a
te
an
s
w
er
i
n
g
o
r
s
elec
tiv
it
y
esti
m
atio
n
.
Ho
w
e
v
er
,
it
co
u
ld
n
o
t
o
b
tai
n
ac
ce
p
tab
le
ap
p
r
o
x
im
a
tio
n
s
f
o
r
th
e
u
n
d
er
l
y
i
n
g
d
ata
s
ets.
2.
E
XI
ST
I
N
G
SYS
T
E
M
Fas
t
R
A
Q
m
eth
o
d
is
p
r
o
p
o
s
ed
in
b
ig
d
ata
en
v
ir
o
n
m
e
n
ts
as
r
ef
er
r
ed
in
[
1
]
.
I
t
f
ir
s
t
d
iv
id
es
lar
g
e
d
ata
in
to
s
m
all
i
n
d
ep
en
d
en
t
p
ar
tit
io
n
s
u
s
i
n
g
a
b
ala
n
ce
d
p
ar
titi
o
n
in
g
al
g
o
r
ith
m
,
an
d
t
h
en
it
g
en
er
ate
s
a
lo
ca
l
esti
m
atio
n
s
k
etc
h
f
o
r
ea
ch
a
n
d
ev
er
y
p
ar
titi
o
n
.
W
h
e
n
a
r
an
g
e
-
a
g
g
r
e
g
ate
q
u
er
y
r
eq
u
e
s
t
ar
r
iv
es,
Fas
t
R
A
Q
o
b
tain
s
t
h
e
r
e
s
u
l
t
f
r
o
m
ea
ch
p
ar
titi
o
n
an
d
it
g
i
v
es
t
h
e
f
i
n
a
l
r
esu
lt
b
y
s
u
m
m
ar
izin
g
th
e
l
o
ca
l
esti
m
ates
f
r
o
m
ea
ch
p
ar
titi
o
n
.
Fas
t
R
A
Q
ap
p
r
o
ac
h
is
i
m
p
le
m
en
ted
o
n
th
e
L
i
n
u
x
p
la
tf
o
r
m
a
n
d
its
p
er
f
o
r
m
an
ce
i
s
ev
al
u
ated
u
s
i
n
g
n
ea
r
l
y
1
0
b
illi
o
n
d
ata
r
ec
o
r
d
s
.
T
h
e
v
ar
io
u
s
al
g
o
r
ith
m
s
u
s
ed
i
n
[
1
]
ar
e
s
tr
atif
icat
io
n
alg
o
r
it
h
m
a
n
d
k
-
m
ea
n
s
al
g
o
r
ith
m
.
T
h
e
d
is
ad
v
an
ta
g
e
s
i
n
[
1
]
ar
e
i
n
ef
f
icie
n
t
r
etr
iev
al
o
f
r
es
u
lt
s
,
s
tr
at
if
ica
t
io
n
p
r
o
b
le
m
o
cc
u
r
s
an
d
m
o
r
e
ti
m
e
co
m
p
lex
it
y
.
T
h
e
f
ast
al
g
o
r
ith
m
s
in
[
3
,
5
]
ar
e
in
tr
o
d
u
ce
d
to
s
p
ee
d
u
p
th
e
r
an
g
e
s
u
m
a
n
d
r
an
g
e
m
ax
q
u
er
ies
i
n
OL
A
P
s
y
s
te
m
.
T
h
e
m
o
s
t
i
m
p
o
r
tan
t
ai
m
o
f
t
h
i
s
s
ce
n
ar
io
is
u
s
ed
to
p
r
e
-
co
m
p
u
te
t
h
e
m
u
lti
-
d
i
m
e
n
s
io
n
al
p
r
ef
i
x
s
u
m
s
o
f
t
h
e
d
ata
c
u
b
e.
T
h
e
to
tal
s
to
r
ag
e
n
ee
d
is
k
ep
t
as
s
a
m
e
as
t
h
e
d
ata
cu
b
e
alo
n
g
w
ith
a
s
m
a
ll
i
n
cr
ea
s
e
i
n
ti
m
e
f
o
r
t
h
e
q
u
er
ie
s
o
f
a
s
i
n
g
leto
n
ce
ll.
Si
n
ce
a
n
y
ce
ll
o
f
th
e
d
ata
c
u
b
e
i
s
ca
lc
u
lated
w
it
h
th
e
s
a
m
e
ti
m
e
co
m
p
le
x
it
y
as
r
a
n
g
e
s
u
m
q
u
er
y
.
T
h
e
ag
g
r
e
g
atio
n
o
p
er
atio
n
s
s
u
c
h
as
m
a
x
,
s
u
m
ar
e
u
s
ed
to
an
s
w
er
u
n
e
x
p
ec
ted
r
u
n
ti
m
e
q
u
er
ie
s
.
I
t
is
ef
f
ici
en
tl
y
u
s
ed
f
o
r
m
a
x
i
m
u
m
r
an
g
e
o
f
q
u
er
ies
u
s
in
g
h
ier
ar
ch
ical
tr
ee
s
tr
u
ct
u
r
e.
T
h
e
d
is
ad
v
an
tag
e
i
s
i
n
f
e
w
ca
s
es it h
as i
s
s
u
e
w
it
h
lo
w
er
ac
cu
r
ac
y
r
ate
s
.
T
h
e
P
r
ef
ix
-
s
u
m
C
u
b
e
(
P
C
)
m
eth
o
d
[
4
,
6
,
7]
is
f
ir
s
t
u
s
ed
i
n
O
L
A
P
to
i
m
p
r
o
v
e
t
h
e
r
a
n
g
e
-
ag
g
r
e
g
ate
q
u
er
y
p
er
f
o
r
m
a
n
ce
.
A
ll
r
a
n
g
e
-
ag
g
r
e
g
ate
q
u
er
ie
s
ar
e
p
r
o
ce
s
s
ed
in
co
n
s
tan
t
ti
m
e
an
d
all
th
e
n
u
m
er
ical
at
tr
ib
u
t
e
v
alu
e
s
ar
e
s
o
r
ted
in
o
r
d
er
,
b
u
t
w
h
en
a
n
e
w
r
o
w
o
r
tu
p
le
i
s
ad
d
ed
in
th
e
cu
b
e,
th
er
e
is
a
n
ee
d
to
r
ec
alcu
late
th
e
p
r
ef
i
x
s
u
m
s
f
o
r
all
d
i
m
e
n
s
i
o
n
s
.
He
n
ce
,
t
h
e
u
p
d
ate
ti
m
e
is
ev
en
w
o
r
s
e.
On
li
n
e
Ag
g
r
e
g
atio
n
(
O
L
A
)
i
s
an
i
m
p
o
r
tan
t
ap
p
r
o
ac
h
to
s
p
ee
d
u
p
r
an
g
e
-
a
g
g
r
eg
ate
q
u
e
r
ies
an
d
is
w
id
el
y
u
s
ed
in
r
elatio
n
al
d
at
ab
ases
[
8
,
9
]
an
d
C
lo
u
d
s
y
s
t
e
m
s
[
1
0
-
1
2
]
.
I
n
OL
A
s
y
s
te
m
s
t
h
e
b
ac
k
g
r
o
u
n
d
co
m
p
u
ti
n
g
p
r
o
ce
s
s
es
r
u
n
f
o
r
a
lo
n
g
ti
m
e.
T
h
e
r
etu
r
n
s
ar
e
r
ef
i
n
ed
an
d
th
e
ac
cu
r
ac
y
is
als
o
g
ettin
g
b
etter
in
s
u
b
s
eq
u
en
t
s
tag
e
s
.
B
u
t
i
n
ea
r
l
y
s
ta
g
es,
t
h
e
u
s
er
s
ca
n
n
o
t
g
et
an
ap
p
r
o
p
r
iate
r
esu
lt
w
it
h
s
atis
f
ied
ac
cu
r
ac
y
.
A
l
s
o
it h
a
v
e
ex
p
e
n
s
i
v
e
s
ce
n
ar
i
o
,
w
asta
g
e
o
f
co
m
p
u
ted
r
eso
u
r
ce
s
,
r
ed
u
ce
d
p
er
f
o
r
m
a
n
ce
[
1
3
-
1
5
]
.
T
h
e
H
y
p
er
L
o
g
L
o
g
al
g
o
r
ith
m
r
ef
er
r
ed
in
[
1
6
]
p
r
o
v
es
to
b
e
ea
s
y
to
co
d
e
a
n
d
e
f
f
ic
ien
t,
b
ein
g
e
v
e
n
n
ea
r
l
y
o
p
ti
m
a
l
u
n
d
er
ce
r
tain
cr
iter
ia.
I
t
is
h
ig
h
l
y
p
r
ac
tical,
v
er
s
atile,
an
d
it
co
n
f
o
r
m
s
well
to
w
h
a
t
an
al
y
s
i
s
p
r
ed
icts
.
T
h
e
alg
o
r
ith
m
u
s
ed
in
[
1
6
]
is
“
H
y
p
er
L
o
g
L
o
g
w
it
h
n
ea
r
o
p
ti
m
a
l
ca
r
d
in
alit
y
al
g
o
r
ith
m
”,
b
u
t
it
h
a
s
th
e
d
is
ad
v
a
n
ta
g
es o
f
lo
w
er
ac
c
u
r
ac
y
a
n
d
s
lo
w
e
x
ec
u
tio
n
ti
m
e
.
A
n
e
w
alg
o
r
it
h
m
r
ep
r
esen
ti
n
g
a
s
er
ies
o
f
i
m
p
r
o
v
e
m
e
n
ts
b
y
r
ed
u
cin
g
th
e
m
e
m
o
r
y
r
eq
u
ir
e
m
en
ts
a
n
d
s
ig
n
i
f
ica
n
tl
y
i
n
cr
ea
s
e
t
h
e
ac
c
u
r
ac
y
f
o
r
an
i
m
p
o
r
tan
t
r
an
g
e
o
f
ca
r
d
in
ali
ties
i
s
u
s
ed
i
n
[
1
7
]
.
I
n
s
in
g
le
p
as
s
,
it
co
m
p
u
tes
t
h
e
lar
g
e
ca
r
d
in
a
liti
es
a
n
d
i
m
p
r
o
v
es
th
e
m
e
m
o
r
y
u
s
a
g
e
m
o
r
e
e
f
f
icien
tl
y
.
T
h
e
al
g
o
r
ith
m
u
s
ed
in
[
1
7
]
is
“
H
y
p
er
lo
g
lo
g
al
g
o
r
ith
m
”.
T
h
e
d
is
ad
v
a
n
ta
g
e
is
it
s
ti
ll h
as a
n
is
s
u
e
w
i
th
er
r
o
r
v
alu
e
s
in
f
e
w
ca
s
e
s
.
3.
O
VE
RVI
E
W
O
F
T
H
E
F
CM
AP
P
RO
ACH
3
.
1
.
P
ro
ble
m
s
t
a
t
e
m
e
nt
T
h
e
ex
is
ti
n
g
s
y
s
te
m
s
s
t
ill
h
as
is
s
u
e
w
it
h
in
e
f
f
icie
n
t
r
etr
ie
v
al
o
f
r
an
g
e
ag
g
r
eg
at
e
q
u
er
ies.
T
h
e
alg
o
r
ith
m
p
r
o
v
id
es
cl
u
s
te
r
in
g
i
n
ac
cu
r
ac
y
f
o
r
lar
g
er
d
ataset.
T
h
e
ex
is
ti
n
g
s
y
s
te
m
lead
s
ti
m
e
co
m
p
le
x
it
y
.
T
h
u
s
th
e
o
v
er
all
s
y
s
te
m
p
er
f
o
r
m
an
ce
i
s
d
eg
r
ad
ed
.
3
.
2
.
K
ey
i
dea
T
h
e
FC
M
ap
p
r
o
ac
h
w
o
r
k
s
i
n
th
e
f
o
llo
w
in
g
m
a
n
n
er
.
First
th
e
b
ig
d
ata
is
d
iv
id
ed
in
to
s
m
a
ll
in
d
ep
en
d
en
t
p
ar
titi
o
n
s
u
s
i
n
g
a
b
alan
ce
d
p
ar
titi
o
n
i
n
g
al
g
o
r
ith
m
as
r
ef
er
r
ed
in
[
1
]
.
Af
ter
t
h
a
t
it
cr
ea
tes
a
lo
ca
l
esti
m
atio
n
s
k
etc
h
f
o
r
ea
ch
i
n
d
ep
en
d
en
t
p
ar
titi
o
n
.
W
h
e
n
a
r
an
g
e
ag
g
r
eg
a
te
q
u
er
y
r
eq
u
est
co
m
es,
F
C
M
ap
p
r
o
ac
h
g
ets
th
e
r
esu
lt
b
y
s
u
m
m
ar
izin
g
lo
ca
l
esti
m
ates
th
at
o
b
tain
ed
f
r
o
m
all
p
ar
titi
o
n
s
.
T
h
e
b
alan
c
ed
p
ar
titi
o
n
in
g
al
g
o
r
ith
m
w
o
r
k
s
alo
n
g
w
it
h
a
p
o
s
ts
tr
atif
ied
s
a
m
p
lin
g
m
o
d
el.
I
t
p
ar
titi
o
n
s
all
th
e
d
ata
in
d
atab
as
e
in
to
d
if
f
er
e
n
t
g
r
o
u
p
s
ac
co
r
d
in
g
to
th
e
ir
attr
ib
u
te
v
al
u
es
a
n
d
it
f
u
r
th
er
s
ep
ar
ates
ea
ch
g
r
o
u
p
in
to
m
u
lt
ip
le
p
ar
titi
o
n
s
w
it
h
r
eg
ar
d
to
th
e
c
u
r
r
en
t d
ata
d
is
tr
ib
u
tio
n
s
a
n
d
th
e
n
u
m
b
er
o
f
s
er
v
er
s
av
ailab
le.
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
A
time
efficien
t a
n
d
a
cc
u
r
a
te
r
etri
ev
a
l o
f ra
n
g
e
a
g
g
r
eg
a
te
q
u
eries
u
s
in
g
fu
z
z
y
clu
s
te
r
in
g
...
(
A
.
Mu
r
u
g
a
n
)
417
T
h
e
esti
m
atio
n
s
k
etc
h
is
a
n
e
w
t
y
p
e
o
f
m
u
lti
-
d
i
m
en
s
io
n
al
h
is
to
g
r
a
m
w
h
ic
h
is
b
u
il
t
w
ith
r
eg
ar
d
to
lear
n
ed
d
ata
d
is
tr
ib
u
tio
n
s
.
T
h
is
m
u
lti
-
d
i
m
e
n
s
io
n
a
l
h
i
s
to
g
r
a
m
ca
n
m
ea
s
u
r
e
t
h
e
q
u
alit
y
o
f
r
o
w
s
o
r
d
ata
s
et
d
is
tr
ib
u
tio
n
s
m
o
r
e
ac
c
u
r
atel
y
.
I
t
m
ai
n
tain
s
al
m
o
s
t
eq
u
iv
ale
n
t
f
r
eq
u
en
cie
s
f
o
r
d
if
f
er
e
n
t
v
alu
es
w
i
th
in
ea
c
h
b
u
ck
et,
e
v
e
n
t
h
o
u
g
h
t
h
e
f
r
e
q
u
en
c
y
d
is
tr
ib
u
tio
n
s
v
ar
y
s
i
g
n
i
f
ica
n
tl
y
.
T
h
is
co
n
ce
p
t
lea
d
s
to
r
ed
u
ce
d
ti
m
e
co
m
p
le
x
it
y
b
y
s
p
litt
in
g
t
h
e
o
v
er
all
f
lo
w
o
f
w
o
r
k
i
n
m
u
ltip
le
p
ar
titi
o
n
s
.
T
h
e
p
ar
titi
o
n
i
n
g
is
d
o
n
e
w
it
h
t
h
e
u
s
e
o
f
attr
ib
u
te
v
al
u
es
t
h
at
co
n
n
ec
ts
m
u
ltip
le
d
atab
ases
.
T
h
is
m
e
ch
an
i
s
m
lead
s
to
f
aster
q
u
er
y
r
esp
o
n
s
e
r
esu
lt
f
o
r
ev
er
y
p
ar
titi
o
n
th
at
co
u
ld
b
e
co
m
b
i
n
ed
later
.
W
h
e
n
a
r
an
g
e
ag
g
r
e
g
ate
q
u
er
y
r
eq
u
est
ar
r
i
v
e
s
,
it
is
d
eli
v
er
ed
in
ea
ch
p
ar
titi
o
n
.
First
th
e
ca
r
d
in
alit
y
e
s
ti
m
ato
r
f
o
r
t
h
e
q
u
e
r
ied
r
an
g
e
i
s
b
u
i
lt
f
r
o
m
t
h
e
h
is
to
g
r
a
m
in
ea
c
h
p
ar
titi
o
n
.
Af
ter
th
at
w
e
esti
m
a
te
th
e
attr
ib
u
te
v
al
u
e
in
ea
ch
p
ar
titi
o
n
,
w
h
ich
is
t
h
e
p
r
o
d
u
ct
o
f
th
e
s
a
m
p
le
an
d
th
e
ca
r
d
in
ali
t
y
v
al
u
e
t
h
at
is
e
s
ti
m
ated
u
s
in
g
t
h
e
est
i
m
a
to
r
.
T
h
e
f
in
al
o
u
tp
u
t
r
e
s
u
l
t
f
o
r
t
h
e
q
u
er
y
r
eq
u
est
i
s
th
e
s
u
m
o
f
all
t
h
e
lo
ca
l e
s
ti
m
at
es f
r
o
m
all
p
ar
titi
o
n
s
.
3
.
2
.
1
.
P
o
s
t
s
t
ra
t
if
ica
t
i
o
n sa
m
pli
ng
P
ar
titi
o
n
in
g
is
a
p
r
o
ce
s
s
o
f
d
i
v
id
in
g
th
e
lar
g
e
tab
le
in
to
m
a
n
y
s
m
all
er
tab
les
b
ased
o
n
th
e
v
alu
e
o
f
a
p
ar
ticu
lar
f
ield
in
a
tab
le
r
ec
o
r
d
.
Data
p
ar
titi
o
n
in
g
is
a
n
i
m
p
o
r
tan
t
s
tep
in
d
ata
an
al
y
s
i
s
b
ec
au
s
e
it
i
m
p
r
o
v
e
s
th
e
q
u
er
y
p
er
f
o
r
m
an
ce
.
I
t
r
e
d
u
ce
s
t
h
e
s
ize
o
f
t
h
e
d
ata
t
o
b
e
s
ca
n
n
ed
f
o
r
th
e
q
u
er
y
r
esu
l
t
a
n
d
h
e
n
ce
th
e
p
er
f
o
r
m
a
n
ce
g
et
s
in
cr
ea
s
e
d
.
I
n
th
e
p
r
o
p
o
s
ed
s
y
s
te
m
p
o
s
t
s
t
r
atif
icatio
n
is
i
n
tr
o
d
u
ce
d
to
o
v
er
co
m
e
t
h
i
s
li
m
itatio
n
w
h
ic
h
w
il
l
s
elec
t
th
e
s
tr
atic
v
ar
iab
le
f
o
r
th
e
e
f
f
i
cien
t d
ata
p
ar
titi
o
n
i
n
g
.
Stra
ti
f
i
ca
tio
n
i
s
s
o
m
eti
m
es
i
n
tr
o
d
u
ce
d
af
ter
t
h
e
s
a
m
p
l
in
g
p
h
a
s
e
in
a
p
r
o
ce
s
s
ca
lled
"
p
o
s
ts
tr
ati
f
icatio
n
"
.
T
h
is
ap
p
r
o
ac
h
is
i
m
p
le
m
e
n
ted
w
h
e
n
t
h
e
ex
p
er
i
m
e
n
ter
d
o
n
o
t
h
av
e
t
h
e
en
o
u
g
h
o
r
n
ec
ess
a
r
y
i
n
f
o
r
m
atio
n
to
cr
ea
te
a
s
tr
atif
y
in
g
v
ar
iab
le
d
u
r
in
g
t
h
e
s
a
m
p
lin
g
p
h
ase.
A
lt
h
o
u
g
h
t
h
e
m
et
h
o
d
is
s
u
s
ce
p
tib
le
to
th
e
p
itf
a
lls
o
f
p
o
s
t
h
o
c
ap
p
r
o
ac
h
es,
it
ca
n
p
r
o
v
id
e
s
ev
er
al
b
en
e
f
it
s
i
n
th
e
r
ig
h
t
s
it
u
atio
n
.
I
m
p
le
m
e
n
t
atio
n
u
s
u
all
y
f
o
llo
w
s
a
s
i
m
p
l
e
r
an
d
o
m
s
a
m
p
le.
P
o
s
ts
tr
ati
f
i
ca
tio
n
ca
n
also
b
e
u
s
ed
to
i
m
p
le
m
e
n
t
w
ei
g
h
t
in
g
,
w
h
ic
h
ca
n
i
m
p
r
o
v
e
th
e
p
r
ec
is
i
o
n
o
f
a
s
a
m
p
le
’
s
esti
m
ates.
P
o
s
ts
tr
atif
icatio
n
i
s
a
m
eth
o
d
f
o
r
ad
j
u
s
ti
n
g
t
h
e
s
a
m
p
li
n
g
w
eig
h
t
s
,
u
s
u
all
y
to
ac
co
u
n
t
f
o
r
u
n
d
er
r
ep
r
esen
ted
g
r
o
u
p
s
in
th
e
p
o
p
u
latio
n
.
P
o
s
ts
tr
ati
f
icatio
n
in
v
o
lv
es
ad
j
u
s
ti
n
g
t
h
e
s
a
m
p
li
n
g
weig
h
ts
s
o
th
at
t
h
e
y
s
u
m
to
th
e
p
o
p
u
latio
n
s
izes
w
it
h
i
n
ea
ch
p
o
s
t
s
tr
atu
m
.
T
h
i
s
u
s
u
all
y
r
es
u
lts
i
n
d
ec
r
ea
s
i
n
g
b
ias
b
ec
au
s
e
o
f
n
o
n
r
esp
o
n
s
e
an
d
u
n
d
er
r
ep
r
esen
ted
g
r
o
u
p
s
in
t
h
e
p
o
p
u
latio
n
.
P
o
s
ts
tr
atif
icatio
n
also
te
n
d
s
to
r
esu
lt
in
s
m
aller
v
ar
ian
ce
esti
m
ate
s
.
T
h
e
s
v
y
s
et
co
m
m
a
n
d
h
as
o
p
tio
n
s
t
o
s
et
v
ar
iab
les
f
o
r
ap
p
ly
in
g
p
o
s
ts
tr
ati
f
icati
o
n
ad
j
u
s
t
m
e
n
ts
to
t
h
e
s
a
m
p
li
n
g
w
eig
h
t
s
.
T
h
e
p
o
s
ts
tr
ata(
)
o
p
tio
n
ta
k
es
a
v
ar
iab
le
t
h
at
c
o
n
tain
s
p
o
s
t
s
tr
atu
m
id
en
ti
f
ier
s
,
a
n
d
th
e
p
o
s
t
w
eig
h
t
(
)
o
p
tio
n
tak
es
a
v
ar
iab
le
t
h
at
co
n
tain
s
t
h
e
p
o
s
ts
tr
atu
m
p
o
p
u
latio
n
s
izes.
I
f
w
j
i
s
th
e
u
n
ad
j
u
s
ted
s
a
m
p
lin
g
w
ei
g
h
t f
o
r
t
h
e
j
th
s
a
m
p
led
i
n
d
iv
id
u
al,
th
e
p
o
s
ts
tr
ati
f
icatio
n
ad
j
u
s
t
ed
s
a
m
p
lin
g
w
ei
g
h
t
is
∑
̂
w
h
er
e
̂
∑
T
h
e
p
o
in
t
esti
m
ate
s
ar
e
co
m
p
u
ted
u
s
i
n
g
th
e
s
e
ad
j
u
s
ted
w
ei
g
h
ts
.
Fo
r
r
ep
licatio
n
-
b
ase
d
v
ar
ian
ce
esti
m
atio
n
,
t
h
e
B
R
R
an
d
j
ac
k
k
n
i
f
e
r
ep
licate
-
w
ei
g
h
t
v
ar
iab
l
es
ar
e
s
i
m
i
lar
l
y
ad
j
u
s
ted
to
p
r
o
d
u
ce
th
e
r
ep
licate
v
alu
e
s
u
s
ed
in
t
h
e
r
esp
ec
ti
v
e
v
ar
ian
ce
f
o
r
m
u
las:
T
h
e
s
co
r
e
v
ar
iab
le
f
o
r
th
e
l
i
n
ea
r
ized
v
ar
ian
ce
e
s
ti
m
ato
r
o
f
a
p
o
s
ts
tr
atif
ied
to
tal
is
(
̂
)
∑
̂
(
̂
̂
)
w
h
er
e
̂
is
th
e
to
tal
esti
m
ato
r
f
o
r
th
e
k
th
p
o
s
ts
tr
atu
m
,
̂
∑
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8708
I
n
t J
E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
10
,
No
.
1
,
Feb
r
u
ar
y
2
0
2
0
:
415
-
420
418
Fo
r
th
e
p
o
s
ts
tr
ati
f
ied
r
atio
esti
m
ato
r
,
th
e
s
co
r
e
v
ar
iab
le
is
(
̂
)
̂
(
̂
)
̂
̂
̂
w
h
er
e
̂
P
is
th
e
p
o
s
ts
tr
atif
ied
to
t
al
esti
m
ato
r
f
o
r
ite
m
x
j
.
3
.
2
.
2
.
F
uzzy
clus
t
er
ing
m
ea
ns
(
F
C
M
)
a
pp
ro
a
ch
I
n
o
u
r
p
r
o
p
o
s
ed
s
y
s
te
m
,
FC
M
clu
s
ter
i
n
g
ca
n
b
e
i
n
tr
o
d
u
c
ed
to
g
r
o
u
p
th
e
s
i
m
ilar
i
n
d
ex
v
alu
e
s
b
y
ca
lcu
lati
n
g
th
e
m
e
m
b
er
s
h
ip
d
eg
r
ee
v
al
u
es
o
f
th
e
m
.
W
ith
t
h
e
h
elp
o
f
FC
M
cl
u
s
ter
in
g
t
h
e
s
i
m
ilar
it
y
b
ased
g
r
o
u
p
in
g
ca
n
b
e
d
o
n
e
ac
cu
r
atel
y
w
h
er
e
th
e
ex
ac
t
s
i
m
ilar
it
y
o
f
c
lass
es
ca
n
b
e
id
en
tif
ied
.
T
h
e
FC
M
clu
s
ter
i
n
g
alg
o
r
ith
m
is
g
i
v
en
a
s
f
o
llo
w
s
:
C
lu
s
ter
i
n
g
o
f
n
u
m
er
ical
d
at
a
f
o
r
m
s
t
h
e
b
asis
o
f
m
a
n
y
class
i
f
icat
io
n
a
n
d
s
y
s
te
m
m
o
d
ell
in
g
alg
o
r
ith
m
s
.
T
h
e
r
ea
s
o
n
f
o
r
cl
u
s
ter
i
n
g
i
s
to
r
ec
o
g
n
ize
t
h
e
n
atu
r
al
g
r
o
u
p
in
g
s
o
f
d
ata
f
r
o
m
a
lar
g
e
d
ata
s
et
in
o
r
d
er
t
o
cr
ea
te
a
c
o
n
cise
r
ep
r
e
s
en
tat
io
n
o
f
a
s
y
s
te
m
’
s
b
eh
a
v
i
o
u
r
.
Fu
zz
y
C
l
u
s
ter
i
n
g
M
ea
n
s
(
FC
M)
is
a
m
et
h
o
d
o
f
cl
u
s
ter
i
n
g
w
h
ich
allo
w
s
o
n
e
p
iece
o
f
d
ata
to
b
elo
n
g
to
t
wo
o
r
m
o
r
e
clu
s
ter
s
.
I
t
is
b
a
s
ed
o
n
m
in
i
m
iza
tio
n
o
f
th
e
f
o
llo
w
i
n
g
o
b
j
ec
tiv
e
f
u
n
cti
o
n
:
∑
∑
W
h
er
e
m
i
s
an
y
r
ea
l
n
u
m
b
er
g
r
ea
ter
th
an
1
,
u
ij
is
th
e
d
eg
r
ee
o
f
m
e
m
b
er
s
h
ip
o
f
x
i
i
n
th
e
c
lu
s
ter
j
,
x
i
is
t
h
e
i
th
o
f
d
-
d
i
m
e
n
s
io
n
al
m
ea
s
u
r
ed
d
ata,
c
j
is
th
e
d
-
d
i
m
e
n
s
io
n
ce
n
tr
e
o
f
th
e
cl
u
s
ter
,
a
n
d
|
|
*
|
|
i
s
an
y
n
o
r
m
e
x
p
r
ess
i
n
g
th
e
s
i
m
ilar
it
y
b
et
w
ee
n
a
n
y
m
e
asu
r
ed
d
ata
a
n
d
th
e
ce
n
tr
e.
Fu
zz
y
p
ar
titi
o
n
in
g
i
s
ca
r
r
ied
o
u
t
th
r
o
u
g
h
an
iter
ati
v
e
o
p
tim
izatio
n
o
f
th
e
o
b
j
ec
tiv
e
f
u
n
ct
io
n
s
h
o
w
n
ab
o
v
e,
w
i
th
t
h
e
u
p
d
ate
o
f
m
e
m
b
er
s
h
ip
u
ij
an
d
th
e
clu
s
ter
ce
n
tr
e
s
c
j
b
y
:
∑
(
)
∑
∑
T
h
is
iter
atio
n
w
il
l
s
to
p
w
h
e
n
{
}
,
w
h
er
e
is
a
ter
m
in
at
io
n
cr
iter
io
n
b
et
w
ee
n
0
an
d
1
,
w
h
er
ea
s
k
ar
e
th
e
iter
atio
n
s
tep
s
.
T
h
is
p
r
o
ce
d
u
r
e
co
n
v
er
g
e
s
to
a
lo
ca
l
m
i
n
i
m
u
m
o
r
a
s
ad
d
le
p
o
in
t o
f
J
m
.
3
.
2
.
3
.
R
a
ng
e
ca
rdina
lity
qu
er
ies
I
n
th
e
ex
i
s
ti
n
g
p
ap
er
[
1
,
18
,
1
9
]
,
it
u
s
es
a
u
n
iq
u
e
R
ec
o
r
d
I
d
to
f
in
d
o
u
t
w
h
et
h
e
r
th
e
ca
r
d
in
alities
th
a
t
ar
e
g
et
f
r
o
m
d
i
f
f
er
en
t
b
u
c
k
et
s
b
elo
n
g
i
n
g
to
th
e
s
a
m
e
r
ec
o
r
d
o
r
n
o
t.
W
e
ca
n
ad
o
p
t
th
e
alg
o
r
ith
m
e
x
p
lain
ed
in
[
1
,
2
0
,
2
1
]
to
esti
m
ate
t
h
e
c
ar
d
in
alit
y
i
n
t
h
e
q
u
er
ies r
an
g
e.
4.
RE
SU
L
T
AND
ANA
L
YS
I
S
A
p
p
r
o
x
i
m
atel
y
2
,
0
0
,
0
0
0
in
p
u
t
d
ata
s
ets
[
22
-
2
7
]
ar
e
u
s
ed
to
f
in
d
th
e
p
er
f
o
r
m
a
n
ce
o
f
FC
M
ap
p
r
o
ac
h
.
T
h
e
s
a
m
e
d
ata
s
et
is
u
s
ed
f
o
r
th
e
e
x
i
s
ti
n
g
s
y
s
te
m
also
a
n
d
its
p
er
f
o
r
m
an
ce
m
ea
s
u
r
es
a
r
e
also
ca
lc
u
lated
.
Af
ter
ap
p
l
y
i
n
g
b
o
t
h
th
e
e
x
i
s
ti
n
g
s
y
s
te
m
an
d
p
r
o
p
o
s
ed
s
y
s
te
m
o
n
t
h
e
s
a
m
e
i
n
p
u
t
d
ata
s
ets
th
e
p
e
r
f
o
r
m
a
n
ce
o
f
p
r
o
p
o
s
ed
FC
M
ap
p
r
o
ac
h
is
b
etter
th
an
t
h
e
ex
i
s
ti
n
g
Fas
t
R
A
Q
ap
p
r
o
ac
h
.
Sa
m
p
le
i
n
p
u
t
d
ata
s
ets
u
s
ed
an
d
th
e
r
esu
ltan
t
g
r
ap
h
s
ar
e
d
es
cr
ib
ed
:
T
h
e
v
ar
io
u
s
p
er
f
o
r
m
an
ce
m
ea
s
u
r
es
ar
e
an
al
y
s
ed
in
o
r
d
er
to
p
r
o
v
e
th
e
p
r
o
p
o
s
ed
s
y
s
te
m
i
s
b
etter
t
h
an
ex
is
ti
n
g
s
y
s
te
m
.
T
h
e
y
ar
e:
i)
ti
m
e
to
e
x
ec
u
te
t
h
e
q
u
er
y
,
i
i
)
ac
cu
r
ac
y
an
d
iii)
er
r
o
r
r
ate.
4
.
1
.
Acc
ura
cy
p
er
f
o
r
m
a
nce
T
h
e
ac
cu
r
ac
y
o
f
F
C
M
ap
p
r
o
ac
h
is
h
i
g
h
er
t
h
an
th
e
ac
c
u
r
ac
y
o
f
all
o
th
er
e
x
is
tin
g
a
p
p
r
o
ac
h
es.
T
h
is
is
s
h
o
w
n
i
n
Fi
g
u
r
e
1.
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
A
time
efficien
t a
n
d
a
cc
u
r
a
te
r
etri
ev
a
l o
f ra
n
g
e
a
g
g
r
eg
a
te
q
u
eries
u
s
in
g
fu
z
z
y
clu
s
te
r
in
g
...
(
A
.
Mu
r
u
g
a
n
)
419
4
.
2
.
E
x
ec
utio
n t
i
m
e
T
h
e
tim
e
ta
k
en
to
ex
ec
u
te
t
h
e
u
s
er
en
ter
ed
r
an
g
e
ag
g
r
eg
a
t
e
q
u
er
y
i
s
s
h
o
w
n
in
F
ig
u
r
e
2
.
T
h
e
tim
e
tak
en
to
e
x
ec
u
te
t
h
e
q
u
er
y
b
y
FC
M
ap
p
r
o
ac
h
is
les
s
th
a
n
t
h
e
ex
is
ti
n
g
ap
p
r
o
ac
h
.
4
.
3
.
E
rr
o
r
r
a
t
e
T
h
e
er
r
o
r
r
ate
g
r
ap
h
f
o
r
th
e
u
s
er
q
u
er
y
is
s
h
o
w
n
in
Fig
u
r
e
3
.
T
h
e
er
r
o
r
r
ate
o
f
FC
M
a
p
p
r
o
ac
h
is
n
ea
r
l
y
5
0
% les
s
th
a
n
t
h
at
o
f
t
h
e
ex
is
ti
n
g
ap
p
r
o
ac
h
er
r
o
r
r
ate.
Fig
u
r
e
1
.
C
o
m
p
ar
is
o
n
o
f
ac
c
u
r
ac
y
Fig
u
r
e
2
.
C
o
m
p
ar
is
o
n
o
f
e
x
ec
u
tio
n
ti
m
e
Fig
u
r
e
3
.
C
o
m
p
ar
is
o
n
o
f
er
r
o
r
r
ate
5.
CO
NCLU
SI
O
NS A
ND
F
UT
URE
E
NH
ANC
E
M
E
NT
I
n
th
is
p
ap
er
,
a
n
e
w
ap
p
r
o
ac
h
Fu
zz
y
C
l
u
s
ter
i
n
g
Me
an
s
(
F
C
M)
ap
p
r
o
ac
h
is
p
r
o
p
o
s
ed
th
at
q
u
ick
l
y
ac
q
u
ir
es
th
e
ac
c
u
r
ate
esti
m
ati
o
n
s
f
o
r
r
an
g
e
-
a
g
g
r
eg
ate
q
u
er
i
es
in
b
ig
d
ata
en
v
ir
o
n
m
e
n
ts
.
Fo
r
ad
-
h
o
c
r
an
g
e
ag
g
r
e
g
ate
q
u
er
ies
F
C
M
h
a
v
e
th
e
ti
m
e
co
m
p
lex
it
y
o
f
O(
n
c
2
p
)
,
w
h
er
e
n
is
n
u
m
b
er
o
f
i
n
s
t
an
ce
s
i
n
th
e
g
i
v
e
n
d
ataset,
c
is
cl
u
s
ter
an
d
p
is
d
ata
p
o
in
ts
.
T
h
i
s
ti
m
e
c
o
m
p
le
x
it
y
i
s
b
etter
th
a
n
t
h
e
p
r
ev
io
u
s
e
x
i
s
ti
n
g
alg
o
r
ith
m
s
.
Fo
r
f
u
t
u
r
e
w
o
r
k
it
is
p
lan
n
ed
to
in
v
esti
g
ate
h
o
w
t
h
is
s
o
lu
tio
n
ca
n
b
e
ex
te
n
d
ed
to
th
e
ca
s
e
m
:
n
f
o
r
m
at
p
r
o
b
le
m
.
Nex
t
it
is
p
l
an
n
ed
to
an
al
y
ze
h
o
w
FC
M
ca
n
b
e
u
s
ed
as
a
to
o
l
to
b
o
o
s
t
th
e
p
er
f
o
r
m
a
n
ce
o
f
d
ata
an
al
y
s
is
i
n
d
atab
ase.
RE
F
E
R
E
NC
E
S
[1
]
X
.
Y
u
n
,
e
t
a
l
.
,
“
F
a
stRA
Q:
A
F
a
st
A
p
p
ro
a
c
h
to
Ra
n
g
e
-
Ag
g
r
e
g
a
te
Qu
e
ru
e
s
in
Big
Da
ta
En
v
iro
n
m
e
n
ts
,
”
IEE
E
T
ra
n
s.
Clo
u
d
Co
m
p
u
t
,
v
o
l.
3
,
2
0
1
5
.
[2
]
P
.
M
ik
a
a
n
d
G
.
T
u
m
m
a
r
e
ll
o
,
“
W
e
b
se
m
a
n
ti
c
s in
th
e
c
lo
u
d
s,”
IE
EE
I
n
tell.
S
y
st.
,
v
o
l
.
2
3
,
p
p
.
8
2
-
8
7
,
2
0
0
8
.
[3
]
H.
Ch
o
i
a
n
d
H.
V
a
rian
,
“
P
re
d
ictin
g
th
e
p
re
se
n
t
w
it
h
G
o
o
g
letre
n
d
s,”
Eco
n
.
Rec
.
,
v
o
l.
8
8
,
p
p
.
2
-
9
,
2
0
1
2
.
[4
]
C.
T
.
Ho
,
e
t
a
l
.
,
“
Ra
n
g
e
q
u
e
ries
i
n
OLA
P
d
a
ta cu
b
e
s,”
ACM
S
IGM
OD
Rec
.
,
v
o
l.
2
6
,
p
p
.
7
3
-
8
8
,
1
9
9
7
.
[5
]
T
.
P
re
is,
e
t
a
l
.
,
“
Qu
a
n
ti
f
y
in
g
trad
in
g
b
e
h
a
v
io
r
in
f
in
a
n
c
ial
m
a
r
k
e
ts
u
sin
g
G
o
o
g
le
tren
d
s,”
S
c
i.
Rep
.
,
v
o
l.
3
,
p
p
.
1
6
8
4
,
2
0
1
3
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8708
I
n
t J
E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
10
,
No
.
1
,
Feb
r
u
ar
y
2
0
2
0
:
415
-
420
420
[6
]
G
.
M
ish
n
e
,
e
t
a
l
.
,
“
F
a
st
d
a
ta
in
th
e
e
ra
o
f
b
ig
d
a
ta:
Tw
it
ter’s
re
a
l
-
ti
m
e
re
late
d
q
u
e
r
y
su
g
g
e
sti
o
n
a
rc
h
it
e
c
tu
re
,
”
Pro
c
.
ACM
S
IGM
OD
In
t.
Co
n
f.
M
a
n
a
g
e
.
Da
t
a
,
p
p
.
1
1
4
7
-
1
1
5
8
,
2
0
1
3
.
[7
]
W
.
L
ian
g
,
e
t
a
l
.
,
“
Ra
n
g
e
q
u
e
ries
i
n
d
y
n
a
m
ic O
LA
P
d
a
ta cu
b
e
s,”
Da
ta
Kn
o
wl.
En
g
.
,
v
o
l.
3
4
,
p
p
.
2
1
-
3
8
,
2
0
0
0
.
[8
]
J.
M
.
He
ll
e
rste
in
,
e
t
a
l
.
,
“
On
li
n
e
a
g
g
re
g
a
ti
o
n
,
”
ACM
S
IGM
OD
Rec
.
,
v
o
l.
2
6
,
p
p
.
1
7
1
-
1
8
2
,
1
9
9
7
.
[9
]
P
.
J.
Ha
a
s
a
n
d
J.
M
.
He
ll
e
rste
in
,
“
Rip
p
le
jo
i
n
s
f
o
r
o
n
li
n
e
a
g
g
re
g
a
ti
o
n
,
”
ACM
S
IGM
OD
Rec
.
,
v
o
l.
2
8
,
p
p
.
2
8
7
-
2
9
8
,
1
9
9
9
.
[1
0
]
E.
Zeitl
e
r
a
n
d
T
.
Risc
h
,
“
M
a
ss
i
v
e
sc
a
le
-
o
u
t
o
f
e
x
p
e
n
siv
e
c
o
n
ti
n
u
o
u
s
q
u
e
ries
,
”
Pro
c
.
V
L
DB
En
d
o
wme
n
t
,
v
o
l.
4
,
p
p
.
1
1
8
1
-
1
1
8
8
,
2
0
1
1
.
[1
1
]
N.
P
a
n
sa
re
,
e
t
a
l
.
,
“
On
li
n
e
a
g
g
re
g
a
ti
o
n
f
o
r
larg
e
M
a
p
Re
d
u
c
e
jo
b
s,”
Pro
c
.
VL
DB
E
n
d
o
wme
n
t
,
v
o
l.
4
,
p
p
.
1
1
3
5
-
1
1
4
5
,
2
0
1
1
.
[1
2
]
T
.
Co
n
d
ie,
e
t
a
l
.
,
“
On
li
n
e
a
g
g
re
g
a
ti
o
n
a
n
d
c
o
n
ti
n
u
o
u
sq
u
e
ry
su
p
p
o
rt
i
n
M
a
p
Re
d
u
c
e
,
”
Pr
o
c
.
A
CM
S
IGM
OD
In
t.
Co
n
f.
M
a
n
a
g
e
.
D
a
ta
,
p
p
.
1
1
1
5
-
1
1
1
8
,
2
0
1
0
.
[1
3
]
Y.
S
h
i,
e
t
a
l
.
,
“
Yo
u
c
a
n
sto
p
e
a
rl
y
w
it
h
c
o
la:
On
li
n
e
p
r
o
c
e
ss
in
g
o
f
a
g
g
r
e
g
a
te
q
u
e
ries
in
th
e
c
lo
u
d
,
”
Pro
c
.
2
1
st
ACM
In
t.
C
o
n
f
.
In
f
.
Kn
o
w.
M
a
n
a
g
e
,
p
p
.
1
2
2
3
-
1
2
3
2
,
2
0
1
2
.
[1
4
]
K.
Bil
a
l,
e
t
a
l
.
,
“
On
th
e
c
h
a
ra
c
ter
iza
ti
o
n
o
f
th
e
stru
c
tu
ra
l
ro
b
u
stn
e
ss
o
f
d
a
ta
c
e
n
tern
e
t
w
o
rk
s,”
IEE
E
T
ra
n
s.
Clo
u
d
Co
mp
u
t
.
,
v
o
l
.
1
,
p
p
.
6
4
-
7
7
,
2
0
1
3
.
[1
5
]
S
.
De
C
.
d
i
V
im
e
rc
a
ti
,
e
t
a
l
.
,
“
In
teg
rit
y
f
o
r
jo
in
q
u
e
ries
in
th
e
c
lo
u
d
,
”
IEE
ET
ra
n
s.
Cl
o
u
d
C
o
mp
u
t
,
v
o
l
.
1
,
p
p
.
1
8
7
-
2
0
0
,
2
0
1
3
.
[1
6
]
S
.
He
u
le,
e
t
a
l
.
,
“
H
y
p
e
rlo
g
l
o
g
in
p
ra
c
ti
c
e
:alg
o
rit
h
m
ic
e
n
g
in
e
e
rin
g
o
f
a
sta
te
o
f
th
e
a
rt
c
a
rd
in
a
li
ty
e
sti
m
a
ti
o
n
a
lg
o
rit
h
m
,
”
Pro
c
.
1
6
th
In
t.
C
o
n
f
.
Exte
n
d
in
g
Da
t
a
b
a
se
T
e
c
h
n
o
l
.
,
p
p
.
6
8
3
-
6
9
2
,
2
0
1
3
.
[1
7
]
P
.
F
laj
o
let,
e
t
a
l
.
,
“
Hy
p
e
rlo
g
lo
g
:
T
h
e
a
n
a
l
y
sis
o
f
a
n
e
a
r
-
o
p
ti
m
a
l
c
a
rd
in
a
li
ty
e
sti
m
a
ti
o
n
a
lg
o
rit
h
m
,
”
Pro
c
.
In
t
.
Co
n
f.
An
a
l
.
Al
g
o
rit
h
ms
,
p
p
.
1
2
7
-
1
4
6
,
2
0
0
8
.
[1
8
]
h
tt
p
:
//
re
se
a
rc
h
.
n
e
u
sta
r.
b
iz/2
0
1
2
/1
2
/1
7
/h
l
li
n
ters
e
c
ti
o
n
s
-
2
/,
2
0
1
2
.
[1
9
]
A
.
T
h
u
so
o
,
e
t
a
l
.
,
“
Hiv
e
a
p
e
tab
y
t
e
sc
a
le
d
a
ta
wa
re
h
o
u
se
u
sin
g
Ha
d
o
o
p
,
”
Pro
c
.
IEE
E
2
6
th
I
n
t.
C
o
n
f.
Da
ta
En
g
.
,
p
p
.
9
9
6
-
1
0
0
5
,
2
0
1
0
.
[2
0
]
T
.
Yu
a
n
d
K.
-
J.
L
in
,
“
A
d
a
p
ti
v
e
a
lg
o
rit
h
m
s
f
o
r
f
in
d
in
g
re
p
lac
e
m
e
n
t
se
rv
ice
s
in
a
u
to
n
o
m
ic
d
ist
rib
u
te
d
b
u
sin
e
ss
p
ro
c
e
ss
e
s,” in
Au
to
n
o
mo
u
s De
c
e
n
tra
li
ze
d
S
y
ste
ms
,
2
0
0
5
.
IS
A
DS
2
0
0
5
.
P
ro
c
e
e
d
i
n
g
s,
p
p
.
4
2
7
–
4
3
4
,
A
p
ril
2
0
0
5
.
[2
1
]
S
.
H.
Ry
u
,
F
.
Ca
sa
ti
,
H.
S
k
o
g
sr
u
d
,
B.
Be
n
a
tallah
,
a
n
d
R.
S
a
i
n
t
-
P
a
u
l,
“
S
u
p
p
o
rti
n
g
th
e
d
y
n
a
m
ic
e
v
o
lu
ti
o
n
o
f
w
e
b
se
rv
ice
p
ro
to
c
o
ls
i
n
se
rv
ice
o
rien
t
e
d
a
rc
h
it
e
c
tu
re
s,”
ACM
T
ra
n
s.
W
e
b
,
v
o
l.
2
,
n
o
.
2
,
p
p
.
1
–
4
6
,
2
0
0
8
.
[2
2
]
D.
M
it
u
z
a
s
,“
P
a
g
e
v
ie
w
sta
ti
stics
f
o
r
w
ik
i
m
e
d
ia
p
ro
jec
ts
,
”
2
0
1
3
.
Av
a
il
a
b
le:
h
tt
p
:
//
d
u
m
p
s.w
i
k
i
m
e
d
ia.o
rg
/o
th
e
r
/p
a
g
e
c
o
u
n
ts
-
ra
w
/
[2
3
]
R
.
Bi,
e
t
a
l
.
,
“
Op
ti
m
izin
g
Re
tran
s
m
issio
n
T
h
re
sh
o
ld
in
W
irele
ss
S
e
n
so
r
Ne
tw
o
rk
s,”
In
d
ia
n
J
o
u
r
n
a
l
o
f
S
c
ien
c
e
a
n
d
T
e
c
h
n
o
l
o
g
y
,
v
o
l
.
16
,
p
p
.
6
6
5
,
2
0
1
6
.
[2
4
]
Ch
risto
p
h
D
o
rn
,
S
c
h
a
h
ra
m
Du
std
a
r,
W
e
ig
h
ted
F
u
z
z
y
Clu
ste
rin
g
f
o
r
Ca
p
a
b
il
it
y
-
d
riv
e
n
S
e
rv
ice
A
g
g
re
g
a
ti
o
n
.
[2
5
]
C.
P
latz
e
r,
F
.
Ro
se
n
b
e
rg
,
a
n
d
S
.
Du
std
a
r,
“
W
e
b
se
r
v
ice
c
lu
ste
rin
g
u
sin
g
m
u
lt
id
im
e
n
sio
n
a
l
a
n
g
les
a
s
p
ro
x
i
m
it
y
m
e
a
su
re
s,”
ACM
T
ra
n
s.
In
ter
n
e
t
T
e
c
h
n
o
l
.
,
v
o
l
.
9
,
n
o
.
3
,
p
p
.
1
–
2
6
,
2
0
0
9
.
[2
6
]
F
.
Ro
se
n
b
e
rg
,
P
.
L
e
it
n
e
r,
A
.
M
i
c
h
lm
a
y
r,
P
.
Ce
li
k
o
v
ic,
a
n
d
S
.
D
u
std
a
r,
“
T
o
wa
rd
s
c
o
m
p
o
siti
o
n
a
s
a
se
rv
i
c
e
-
a
q
u
a
li
ty
o
f
se
rv
ic
e
d
riv
e
n
a
p
p
ro
a
c
h
,
”
2
9
2
0
0
9
-
A
p
ril
2
2
0
0
9
,
p
p
.
1
7
3
3
–
1
7
4
0
,
2
0
0
9
.
[2
7
]
E.
M
a
x
im
il
ien
a
n
d
M
.
S
i
n
g
h
,
“
S
e
lf
-
a
d
ju
stin
g
tru
st a
n
d
se
lec
ti
o
n
f
o
r
w
e
b
se
rv
i
c
e
s,”
p
p
.
3
8
5
–
3
8
6
,
Ju
n
e
2
0
0
5
.
Evaluation Warning : The document was created with Spire.PDF for Python.