Internati
o
nal
Journal of Ele
c
trical
and Computer
Engineering
(IJE
CE)
Vol.
5, No. 6, Decem
ber
2015, pp. 1564~
1
568
I
S
SN
: 208
8-8
7
0
8
1
564
Jo
urn
a
l
h
o
me
pa
ge
: h
ttp
://iaesjo
u
r
na
l.com/
o
n
lin
e/ind
e
x.ph
p
/
IJECE
Clutter Reduction in Paralle
l Coordinates using Binning
Approach for Improved Visualization
Swathy Sunil Kum
a
r,
Teen
u Krishnan, Sreeja As
hok,
M.
V.
Jud
y
Department of
Computer Scien
ce
& I
.
T,
Amrita School of
Arts
& Sciences
, Kochi,
Amrita
Vishwa Vidy
apeetham
Article Info
A
B
STRAC
T
Article histo
r
y:
Received Apr 23, 2015
Rev
i
sed
Au
g
11
, 20
15
Accepted Aug 30, 2015
As the data and number of in
forma
tion sources keeps on mounting, th
e
m
i
ning of neces
s
a
r
y
in
form
ation
and the
i
r pres
e
n
tation
in a hu
m
a
n delic
at
e
form
becom
e
s
a great ch
all
e
n
g
e. Vis
u
ali
zat
io
n helps
us
to pictori
a
l
l
y
represent, evalu
a
te and
un
cov
e
r
th
e knowledge from the
data u
nder
consideration. Data visual
ization
offers its immense
opportunity
in the fields
of trad
e, b
a
nkin
g
, fin
a
nce,
insurance, en
erg
y
etc. With th
e d
a
ta
explosion in
various
fields
, t
h
ere is
a larg
e i
m
porta
nce for visualization tech
niques. Bu
t
when the quan
t
ity
of d
a
ta b
eco
mes el
evated, th
e visualization
methods may
take awa
y
th
e co
m
p
etenc
y
. P
a
r
a
ll
el coordin
a
tes
is
an em
inent and often us
ed
m
e
thod for data
visualizat
ion. H
o
wever
the efficiency
of th
is m
e
thod will be
abridged if ther
e are large amount of inst
ances in the dataset, th
ereb
y
making
the vis
u
al
iz
ation
clum
s
i
er and t
h
e dat
a
retr
ieva
l
ver
y
ineffi
ci
ent
.
Here we
introduced a d
a
ta summarizatio
n approach
as
a prepro
cessing
step to
the
existing
parallel coord
i
nate
method
to make th
e visu
alization mor
e
proficient.
Keyword:
B
i
nni
n
g
Data v
i
su
alizatio
n
tech
n
i
qu
es
p
a
rallel coord
i
n
a
tes
Mu
ltiv
ariate d
a
ta
Param
e
tric m
e
t
h
od
s
Copyright ©
201
5 Institut
e
o
f
Ad
vanced
Engin
eer
ing and S
c
i
e
nce.
All rights re
se
rve
d
.
Co
rresp
ond
i
ng
Autho
r
:
Sr
eej
a
A
s
hok
,
Depa
rt
m
e
nt
of
C
o
m
put
er Sci
e
nce &
I
.
T,
Am
rita Sch
o
o
l
o
f
Arts & Scien
ces,
Am
rita Vishwa
Vidy
a
p
eetham
.
Em
a
il: sreej
a.ash
ok@g
m
ail.c
o
m
1.
INTRODUCTION
Data
v
i
su
alizatio
n techn
i
qu
es
o
f
ten
p
l
ay a
v
ital ro
le wh
ile rep
r
esen
ting
large qu
an
tities o
f
d
a
ta,
h
e
l
p
anal
y
s
i
ng t
h
es
e dat
a
an
d l
a
n
d
i
n
g at
co
nvi
n
c
i
ng c
oncl
u
si
o
n
s [
3
]
.
Dat
a
vi
sual
i
zat
i
on,
w
h
en
d
one
faul
t
l
essl
y
,
p
r
ov
es to
b
e
an
efficien
t m
ean
s t
o
co
m
p
reh
e
nd
t
h
e
d
a
ta
th
at are bu
ried in
larg
e d
a
tasets to
d
i
scov
er th
e
relationships, correla
tions, outliers and
hidde
n patte
rns. It is accepte
d as the
m
o
st efficient
m
e
t
h
od for
co
nv
eying
t
h
e
in
fo
rm
atio
n
in
th
e
d
a
taset to
t
h
e i
n
tend
ed
us
ers with
t
h
e he
lp of graphical
aids suc
h
a
s
t
a
bles
and c
h
art
s
.
Pr
o
cessi
ng
, anal
y
z
i
ng an
d c
o
m
m
uni
cat
i
n
g t
h
e d
a
t
a
resi
di
n
g
i
n
l
a
rge dat
a
set
s
prese
n
t
a vari
e
t
y
of
ethical and a
n
a
l
ytical challe
nges f
o
r
dat
a
vi
su
al
i
zat
i
on.
Data v
i
su
alizatio
n
fi
nd
s its
ap
p
licab
ility in
d
i
fferen
t
sph
e
res su
ch
as ch
em
ical i
m
a
g
ing
,
crim
e
m
a
ppi
n
g
,
bi
ol
ogi
cal
dat
a
vi
sual
i
zat
i
on,
m
e
di
cal
im
agi
n
g a
n
d
so
o
n
.
C
h
em
i
c
al
im
agi
n
g i
s
a
n
a
n
al
y
t
i
c
al
capability of c
r
eating a
visua
l
im
ages
of com
ponents
distribu
tion
from
conc
urre
nt m
e
asurem
ent of s
p
ectral,
sp
atial an
d
time in
fo
rm
atio
n
.
Med
i
cal i
m
ag
i
n
g
is t
h
e prac
tice o
f
creating
v
i
su
al illu
stratio
n
of th
e in
terio
r
o
f
a
bo
dy
f
o
r cl
i
n
i
c
al
st
udy
an
d m
e
di
cal
i
n
t
e
rfe
re
nces.
U
nde
rs
tand
ing
th
e
raw
data wh
ich is resid
i
ng
in
th
e l
a
rge
dat
a
set
i
s
a
n
i
m
port
a
nt
, y
e
t
chal
l
e
ngi
n
g
di
l
e
m
m
a i
n
t
h
e c
u
r
r
ent
si
t
u
at
i
o
ns
w
h
ere a
l
a
r
g
e am
ount
dat
a
get
s
accum
u
lated every
w
he
re. As the qua
ntity
of records incre
a
ses,
the effectiveness
with
which the
data can be
interpreted re
duces. T
h
ere are
m
a
ny vi
sual
i
zat
i
on t
ech
ni
q
u
es t
h
at
can be
used t
o
vi
s
u
al
i
ze hi
gh di
m
e
nsi
onal
dat
a
s
u
ch
as sc
at
t
e
r pl
ot
,
gl
y
p
h
s,
pa
ral
l
e
l
co
o
r
di
nat
e
s,
hi
era
r
chi
cal
Tec
hni
que
s [
3
]
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
J
ECE
I
S
SN
:
208
8-8
7
0
8
Clu
tter
Redu
ctio
n
in
Pa
ra
llel Co
o
r
d
i
na
tes u
s
in
g
Binn
ing
App
r
oa
ch
f
o
r Impro
v
ed
…
(S
wa
t
h
y S
u
n
il
Kumar)
1
565
Scat
t
e
r pl
ot
s ar
e t
h
e el
derl
y
and
gene
ral
l
y
used m
e
t
hod t
o
pr
o
j
ect
hi
g
h
di
m
e
nsi
onal
dat
a
i
n
t
o
a t
w
o
dim
e
nsional space. In t
h
is, t
h
e
pa
rallel proj
ec
tions a
r
e
po
sitione
d i
n
gri
d
structure to aid t
h
e use
r
to m
e
morize
the dim
e
nsions related with each pr
ojecti
o
n. Glyphs are graphical obj
ects that are
designe
d to
convey
m
u
l
tip
le d
a
ta v
a
lu
es. Th
is tech
n
i
q
u
e
can
b
e
u
s
ed
on
ly wh
en
th
ere is a limited
n
u
m
b
e
r of a d
a
ta ele
m
en
t to
b
e
di
spl
a
y
e
d
si
m
u
l
t
a
neou
sl
y
,
as i
t
m
a
y
requi
r
e
a
l
a
rge
am
ount
of
scree
n
s
p
ac
e t
o
be
vi
ewe
d
.
Paral
l
e
l
co
o
r
d
i
nat
e
s
are
p
r
in
ci
p
a
lly p
opu
lar t
o
d
a
y
d
u
e
to
its t
h
eoretical si
m
p
lici
t
y
an
d so
lid
ap
pearan
ce.
Paral
l
e
l
coo
r
di
nat
e
s are hi
gh
-
d
i
m
ensi
onal
da
t
a
vi
sual
i
zat
i
o
n
t
echni
q
u
e w
h
i
c
h was i
n
ve
nt
e
d
i
n
1
9
8
0
’
s
that represe
n
ts
N-dim
e
nsiona
l data
in a
2-dim
e
nsional s
p
ace wit
h
m
a
them
atical rigorous
ne
ss [2], [20]. It
find
s its ap
p
l
i
cab
ility in
d
i
v
e
rse sets of mu
ltid
i
m
en
sio
n
a
l p
r
ob
lem
s
in
m
a
n
y
d
o
m
ai
n
s
, su
ch
as
W
i
nViz,
Xm
dvTo
o
l
,
a
n
d SP
SS
Di
am
ond
[
12]
,
[1
7]
.
I
t
uses
paral
l
e
l
axi
s
f
o
r
di
m
e
nsi
ons a
n
d re
p
r
e
s
ent
s
N
di
m
e
nsi
onal
dat
a
i
n
t
w
o
di
m
e
nsi
onal
s
p
a
ces.I
dent
i
f
y
i
n
g
t
h
e cl
ust
e
rs i
n
t
h
e pl
ot
s i
s
a
n
im
port
a
nt
pa
rt
of
u
n
d
erst
a
ndi
ng
an
d
in
terpreting
the d
a
ta [4
],
[8
]. Figu
re
1
represen
ts th
e p
a
ral
l
e
l
coo
r
di
nat
e
p
l
ot
of
a
dat
a
se
t
whi
c
h c
onsi
t
s
of
5
at
t
r
i
but
es
a
n
d 4 dat
a
o
b
j
ect
s.
Fi
gu
re 1.
Paral
l
el
coo
r
di
nat
e
pl
ot
, 5
at
t
r
i
but
e
s
an
d 4 dat
a
ob
ject
s
In the
rece
nt years m
a
ny research e
f
forts
ha
ve bee
n
d
i
rected
at th
e d
i
sp
lay o
f
larg
e d
a
ta sets as well
as i
n
t
h
e a
r
ea
of
pa
ral
l
e
l
coo
r
di
nat
e
s.
Di
ffe
rent
e
nha
nc
em
ent
t
echni
q
u
es ca
n
be i
n
-co
o
p
erat
ed
w
i
t
h
t
h
e
paral
l
e
l
co
o
r
di
nat
e
f
o
r
bet
t
e
r an
d i
m
pro
v
ed
dat
a
vi
sua
lizatio
n
.
U
s
e of
co
lor
s
, an
im
ati
o
n,3D
v
i
ew
i
n
g, and
br
us
hes, al
l
ca
n hel
p
us i
n
t
h
e
bet
t
e
r u
n
d
erst
andi
ng
o
f
t
h
e
d
a
t
a
[2]
,
[
1
1]
. P
a
ral
l
e
l
coo
r
di
n
a
t
e
pl
ot
, acc
om
pani
e
d
b
y
scatter p
l
o
t
an
d th
e
rad
a
r ch
art
h
a
s
b
een ex
ten
s
i
v
ely
worn
fo
r v
i
su
alizing
m
u
ltiv
ariate d
a
tasets [1
]. Th
e
u
s
e
of c
o
lors and
opacity can e
n
hance t
h
e vis
u
alization.
Hi
g
h
l
i
ght
i
ng t
h
e sp
eci
fi
c dat
a
can
im
prove t
h
e
vi
sual
un
de
rst
a
n
d
i
n
g
and
t
h
e
r
eby
i
n
crease t
h
e
vi
su
al
cl
ari
t
y
. Al
l
e
n R
M
a
rt
i
n
st
u
d
i
e
d t
h
e
hi
g
h
d
i
m
e
nsi
onal
br
u
s
hi
n
g
o
f
m
u
ltiv
ariate d
a
ta, it is
an
o
p
e
ratio
n
est
a
b
lish
e
d
in
the v
i
su
alizatio
n syste
m
s
to
in
teractiv
ely sel
ect th
e
subsets from
the original dat
a
set.
He desc
ribed
N-dim
e
nsi
onal brushe
s whic
h are de
fined
on
data space. It
pr
o
v
i
d
es ad
va
n
c
em
ent
t
o
t
h
e Xm
dv To
ol
wh
i
c
h can be
use
d
f
o
r dat
a
vi
su
al
i
zat
i
on by
pr
ovi
di
n
g
a hi
g
h
l
i
ght
i
n
g
ope
rat
i
o
n t
o
t
h
e use
r
by
usi
n
g a
si
n
g
l
e
b
r
us
h
[1
8]
.
It
al
so
pr
o
v
i
d
e
d
Xm
dv t
o
ol
wi
t
h
a
r
a
nge
o
f
m
e
t
h
o
d
s
of
br
us
h speci
fi
ca
t
i
on as wel
l
as m
a
ni
pul
at
i
o
n.
Jim
m
y
Johans
s
on st
udi
e
d
di
f
f
e
rent
m
e
t
hods
i
n
whi
c
h a dat
a
t
upl
e
with n dim
e
nsion ca
n be re
presented as polylines connectin
g n points [20]. It is a space efficient as well as an
in
teractiv
e m
e
t
h
od
to
rep
r
esen
t a larg
e
d
a
ta
set [9
]. He
re cl
u
s
tering
algorith
m
is u
s
ed
in
co
m
b
in
atio
n
wit
h
th
e
paral
l
e
l
co
or
d
i
nat
e
’s m
e
t
hodol
ogy
t
o
rep
r
esent
t
h
e
dat
a
. Hi
g
h
p
r
eci
si
on t
e
xt
ures
are use
d
f
o
r
bet
t
e
r
v
i
su
alizatio
n an
d th
e cl
u
s
ters
are
h
i
gh
lig
h
t
ed in
d
i
fferen
t
colo
rs.
Parallel co
ord
i
n
a
tes h
a
v
e
b
e
en
prov
ed
to
b
e
an
efficien
t tool d
u
e
to
its effi
cien
cy in
po
in
t
i
n
g
o
u
t
th
e
si
m
ilarit
y
b
e
tween
each
attribu
t
es,
bu
t efficiency re
duces
due to polylines
an
d
ove
r
pl
ot
t
i
ng
[
7
]
.
D
u
e t
o
t
h
e
clu
tter im
p
r
essio
n
and
i
n
terferen
ce
with
cro
ssi
n
g
lin
es
,
th
e op
eratio
n
su
ch
as
selectio
n
as
well as d
a
ta
clu
s
tering
b
e
comes a ch
allen
g
i
n
g
prob
lem
w
ith
resp
ect to
larg
e d
a
taset.The p
r
esen
ce o
f
po
lylin
es redu
ces th
e
v
i
sib
ility o
f
h
i
d
d
e
n
p
a
ttern
s
in
larg
e
d
a
ta set. Data red
u
ctio
n
techn
i
ques i
m
p
r
ov
e the v
i
su
alizatio
n and
deci
si
o
n
m
a
ki
ng
by
ret
a
i
n
i
n
g t
h
e
pai
r
wi
s
e
cor
r
el
at
i
o
n
bet
w
ee
n eac
h
at
t
r
i
but
e
val
u
e
s
[
13]
.
Her
e
we are
pr
o
posi
n
g
a
da
t
a
cent
r
i
c
a
p
p
r
oach
f
o
r
dat
a
s
u
m
m
a
ri
zat
i
on to re
duce the
effect of
clutters
where t
h
e
groupi
ng
i
s
do
ne ah
ead
of
pat
t
e
rn
ge
ne
rat
i
o
n
.
Thi
s
i
m
pr
o
v
es t
h
e c
o
m
p
ari
s
on o
f
i
n
di
vi
d
u
al
cha
r
a
c
t
e
ri
st
i
c
s of ea
ch dat
a
o
b
j
ect
with
resp
ect to
t
h
e co
mp
lete d
a
ta set.
2.
R
E
SEARC
H M
ETHOD
In t
h
e p
r
o
p
o
se
d m
e
t
hodol
o
g
y
we at
t
e
m
p
t
t
o
red
u
ce t
h
e
sh
or
t
c
om
i
ngs
of
pa
ral
l
e
l
coo
r
di
nat
e
pri
m
ari
l
y
o
v
e
r
p
l
o
tting
,
b
y
co
m
b
in
in
g
b
i
nn
ing
m
e
t
h
od
s
with
p
a
rallel co
o
r
d
i
n
a
t
e
s to
b
o
o
s
t t
h
e efficien
cy
of d
a
ta
in
terpretatio
n.
Th
is is ach
ieved
b
y
ad
d
i
ng
b
i
nn
ing
as
a
preproce
ssing st
ep to t
h
e norm
al parallel coordi
nate
app
r
oach
.
Fi
gu
re 2 depi
ct
s
t
h
e
w
o
r
k
fl
o
w
of
t
h
e pr
o
pose
d
sy
st
em
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
SN
:
2
088
-87
08
IJECE
Vol. 5, No. 6, D
ecem
ber
2015 :
1564 –
1568
1
566
Fi
gu
re 2.
Wor
k
fl
o
w
of
t
h
e p
r
op
ose
d
sy
st
em
The following steps
a
r
e
involv
ed in t
h
e
propose
d system
1)
Data Norm
al
iz
atio
n
2)
B
i
nni
n
g
usi
n
g param
e
t
r
i
c
m
odel
s
3)
Visu
alizatio
n usin
g p
a
rallel coo
r
d
i
n
a
tes
2.
1.
D
a
t
a
N
o
r
m
al
i
z
ati
o
n
To a
v
oi
d
de
p
e
nde
nce
o
n
t
h
e sel
ect
i
o
n
o
f
m
easurem
ent u
n
i
t
,
t
h
e
dat
a
sh
o
u
l
d
be
n
o
r
m
a
li
zed. It
in
vo
lv
es th
e l
i
n
ear transfo
r
matio
n
o
f
d
a
t
a
to
fall
with
in
a co
mm
o
n
ran
g
e. Thro
ug
h
n
o
rm
aliza
t
i
o
n, all
attrib
u
t
es will
b
e
g
i
v
e
n
an
eq
u
a
l weigh
t
. Man
y
no
rm
ali
zatio
n
m
e
th
o
d
s are av
ailab
l
e su
ch
as m
i
n
m
a
x
no
rm
al
i
z
at
i
on, Z-sc
ore n
o
rm
al
i
zat
i
on, deci
m
a
l
scal
i
ng et
c.Here m
i
n–
m
a
x norm
a
l
i
zat
i
on
m
e
t
hod i
s
bei
n
g use
d
to
g
e
t a
p
o
sitiv
e rang
e of
valu
es with
i
n
a li
m
i
t to
o
b
s
erv
e
t
h
e v
a
rian
ces
o
f
each
attrib
u
t
e clearl
y
. Th
e
m
i
nim
u
m
and
m
a
xim
u
m
val
u
es are
set
as
0
and
1
.
S
o
t
h
i
s
m
a
kes i
t
i
n
t
o
a
no
n
ne
gat
i
v
e
r
a
nge
.
Z-sc
ore
norm
alization is formulated as:
M
i
n m
a
x no
rm
al
i
zat
i
on i
s
f
o
r
m
ul
at
ed as
gi
v
e
n
bel
o
w.
2.2. Binning using P
a
r
a
metr
i
c Appr
oaches
By reduci
ng t
h
e size of the
da
taset, the cluttering ef
fect can
be re
duced. One approach is t
o
us
e data
di
scret
i
zat
i
on
by
gr
ou
pi
n
g
d
a
t
a
object
s i
n
t
o
i
n
t
e
rval
s
.
B
i
nni
ng i
s
a t
o
p
-
d
o
w
n
t
eari
n
g
t
echni
q
u
e bas
e
d o
n
preci
se
n
u
m
b
er o
f
bi
ns
. T
h
ere are
di
ffe
r
e
nt
t
y
pes
of
bi
n
n
i
n
g a
p
p
r
o
aches:
e
qual
wi
dt
h
bi
nni
ng
,
eq
ual
fre
que
ncy
bi
n
n
i
ng.
T
h
e
pr
o
p
o
s
ed sy
st
em
use
s
eq
ual
wi
dt
h
bi
n
n
i
n
g t
e
c
hni
que
. T
h
e
m
a
i
n
chal
l
e
n
g
e
he
r
e
i
s
t
o
i
d
ent
i
f
y
t
h
e
o
p
t
i
m
u
m
bi
n w
i
dt
h an
d
bi
n
n
u
m
b
er. St
u
r
ge
s'
form
ul
a for
no
rm
al
di
st
ribut
i
o
n i
s
a st
a
t
i
s
t
i
cal
pr
oce
d
u
r
e
fo
r
deci
di
n
g
t
h
e
o
p
t
i
m
u
m
bi
n si
ze w
h
i
c
h i
s
m
o
re
ef
fi
ci
ent
w
h
en
com
p
are
d
t
o
ot
her
m
e
t
hods l
i
k
e
ri
sk m
i
nim
i
zat
ion t
e
c
hni
que a
nd B
a
y
e
si
an
o
p
t
i
m
a
l
bi
nni
n
g
[2
2]
. Acc
o
r
d
i
n
g t
o
St
u
r
ge
s'
t
h
eory
, t
h
e
o
p
t
i
m
a
l
bi
n
si
ze k i
s
de
ri
ve
d
usi
n
g t
h
e f
o
r
m
ul
a.
Whe
r
e
‘n’ is t
h
e size of
data
Evaluation Warning : The document was created with Spire.PDF for Python.
I
J
ECE
I
S
SN
:
208
8-8
7
0
8
Clu
tter
Redu
ctio
n
in
Pa
ra
llel Co
o
r
d
i
na
tes u
s
in
g
Binn
ing
App
r
oa
ch
f
o
r Impro
v
ed
…
(S
wa
t
h
y S
u
n
il
Kumar)
1
567
Th
e
nu
m
b
er
of b
i
ns,
k
is d
eci
d
e
d b
a
sed on th
e size
of th
e
d
a
tab
a
se. Th
e
ex
istin
g d
a
taset is rep
l
aced
b
y
sm
o
o
t
h
i
ng
u
s
ing
b
i
n
b
ound
ar
ies an
d th
e
n
e
w
l
y g
e
n
e
r
a
ted
d
a
taset is
g
i
ven
as inpu
t fo
r v
i
su
alizatio
n
u
s
ing
p
a
rallel co
ord
i
n
a
tes. Th
e
b
i
nn
ing
app
r
o
a
ch
p
a
rtitio
n
s
t
h
e dataset in
to
su
itab
l
e b
i
n
size to av
o
i
d
cl
u
ttering
and
sim
p
lifies the
dataset. T
h
e
bi
ns t
hus
create
d
are
plo
tted which
bring
s
m
o
re clarity for
furth
e
r
p
r
o
cess.
2
.
3
.
Visua
liza
t
io
n
using
Pa
ra
llel
Coo
r
dinates
The i
n
p
u
t
dat
a
set
can
be
si
m
p
l
i
f
i
e
d a
n
d e
x
e
c
ut
ed
ve
ry
fast
aft
e
r
t
h
e
bi
n
n
i
n
g
p
r
ocess.
B
i
nne
d
pa
ral
l
e
l
coo
r
di
nat
e
s p
r
ovi
des co
nt
ext
vi
ews o
f
t
h
e d
a
t
a
set
rat
h
er
than the foc
u
s vi
ews. T
h
e cl
utters can be effe
ctively
red
u
ce
d t
h
r
o
u
g
h
whi
c
h
we ca
n easi
l
y
di
st
i
n
gui
s
h
t
h
e
pat
t
e
rns
.
Eac
h
at
t
r
i
but
e i
s
re
pre
s
e
n
t
e
d
usi
n
g
a
v
e
rt
i
cal
l
i
n
e and t
h
e bi
n sam
p
l
e
s are hi
g
h
l
i
ght
e
d
as
ho
ri
zo
nt
al
lines. The
use
r
rec
e
ives i
mmediate feedbac
k
about the
characte
r
istics of the
data
objects an
d the
c
o
rrelation bet
w
een each attribut
es.
The pair wise
com
p
ari
s
on of
each data and com
p
arison of varia
b
les
associated with e
ach data ite
m
is
clearly differentiated usi
n
g this
pr
ocess
.
Dat
a
c
l
ust
e
rs a
ppea
r
as den
s
e re
gi
o
n
s w
h
i
c
h
sh
o
w
the sim
ilarity
of
features
ass
o
ciated with ea
ch data
ite
m
.
3.
RESULT AND DIS
C
USSI
ON
W
e
e
x
am
i
n
ed t
h
e effect
i
v
e
n
ess o
f
t
h
e
p
r
o
p
o
sed a
p
pr
o
ach t
h
r
o
ug
h e
xpe
ri
m
e
nt
s on
di
ffe
rent
d
a
tab
a
ses su
ch as Data_User_
Mod
e
ling
_
Dataset_
Ha
m
d
i_
To
lg
a
KA
H
R
A
M
AN
h
a
v
i
ng
258
in
stan
ces and
6
at
t
r
i
but
es, Si
t
k
a89
havi
ng
63
2 i
n
st
ances a
n
d 4 at
t
r
i
but
e,
IR
IS ha
vi
n
g
1
50 i
n
st
a
n
ces a
nd
5at
t
r
i
b
ut
es.
The
num
ber o
f
bi
n
s
depe
n
d
s o
n
t
h
e n
u
m
b
er of i
n
st
ances i
n
t
h
e
dat
a
base.
N
u
m
b
er of bi
n
s
are cal
cul
a
t
e
d b
a
sed o
n
St
ur
ges
’
f
o
rm
ul
a .The
si
m
u
l
a
ti
on
was
d
o
n
e
u
s
i
n
g
R
p
r
og
ra
m
m
i
ng l
a
ng
ua
ge.
The
pr
op
ose
d
m
e
t
hod
pr
o
v
i
d
es m
o
re cl
ari
t
y
and
u
n
d
erst
a
n
di
n
g
o
f
t
h
e
dat
a
set
.
The c
o
nv
erge
nces a
r
e
m
o
re accurate
and clear t
o
study the influence
of
each
attribute
on the outcom
e. The pa
rallel coordinate
represe
n
tation
of each
dataset before an
d aft
e
r applying the
binni
ng approach
is shown i
n
Figure 3, 4 and
5.
Th
e left im
ag
e represen
ts th
e
trad
itio
n
a
l lin
e b
a
sed
p
a
rallel co
ord
i
n
a
tes and
ri
g
h
t
im
ag
e sh
ows
b
i
nn
ing
b
a
sed
paral
l
e
l
co
o
r
di
nat
e
s
wi
t
h
9 ,
1
0 a
n
d
8
bi
ns
pe
r
dat
a
di
m
e
nsi
o
n
ba
sed
o
n
t
h
e dat
a
si
ze.
Fi
gu
re
3.
C
o
m
p
ari
s
on
o
f
t
w
o
Paral
l
e
l
co
or
di
nat
e
re
n
d
eri
ngs
i
n
t
h
e
sam
e
dat
a
set
Data_
U
ser
_
M
odelin
g
_
Datas
e
t_Ham
d
i_Tol
g
a KA
HRAM
AN
Fig
u
re
4
.
Dem
o
n
s
t
r
ate th
e
app
earan
ce
o
f
data set,
sitk
a8
9 usin
g trad
ition
a
l
(left) and
b
i
nned
p
a
rallel
coo
r
di
nat
e
sy
st
em
Fig
u
re
5
.
Illu
st
ratin
g th
e
p
a
ttern
s av
ailab
l
e in IRIS dataset
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
SN
:
2
088
-87
08
IJECE
Vol. 5, No. 6, D
ecem
ber
2015 :
1564 –
1568
1
568
Disp
lay of th
e d
a
ta v
a
l
u
es
usin
g
b
i
nn
ing
ap
pro
ach
im
p
r
ov
es
p
a
ttern
opti
m
izatio
n
.
By co
m
b
in
in
g
b
i
nn
ing
an
d p
a
rallel co
ord
i
n
a
t
e
s, th
e cl
u
tters are
redu
ced
d
r
astically. Th
is i
m
p
r
ov
es the
perform
a
nce and the
effect
i
v
e
n
ess
o
f
deci
si
on
m
a
ki
ng
fr
om
t
h
e l
a
rge
dat
a
set
s
4.
CO
NCL
USI
O
N
Thi
s
pa
per f
o
c
u
se
d on
red
u
ci
ng t
h
e
ove
r pl
ot
t
i
ng i
n
pa
ral
l
el
coor
di
nat
e
s
by
redu
ci
n
g
t
h
e cl
ut
t
e
rs
usi
n
g
bi
n
n
i
n
g
app
r
oach
. T
h
e
i
n
co
rp
o
r
at
i
o
n
of
bi
n
n
i
n
g m
e
tho
d
o
l
o
gy
i
n
t
o
t
h
e pa
ral
l
e
l
co
or
di
nat
e
pl
ot
a
ssi
st
s
b
und
lin
g
o
f
b
i
n
n
e
d
d
a
ta,
wh
i
c
h
i
m
p
r
ov
es the p
e
rcep
tib
ility
o
f
d
a
ta wh
en
co
m
p
ared
to
the o
r
ig
i
n
al p
l
o
t
. The
pl
ot
s
obt
ai
ne
d
aft
e
r i
n
c
o
r
p
ora
t
i
ng t
h
e
pr
o
p
o
s
ed m
e
t
hod
ol
o
g
y
su
p
p
o
r
t
i
t
s
users
t
o
ar
ri
ve
at
val
i
d
co
ncl
u
si
o
n
s
fro
m
th
e larg
e d
a
tasets. Th
e v
i
su
alization
b
eco
m
e
s
m
o
re v
a
lu
ab
le an
d p
a
tterns become
m
o
re d
e
tectab
le,
t
h
ere
b
y
i
m
prov
i
ng t
h
e e
ffec
tiveness of visual
ization.
ACKNOWLE
DGE
M
ENTS
Th
is work
is su
ppo
rted
b
y
th
e
D
S
T Funded
Pr
oj
ect,
(
S
R
/
C
SI/
81/
20
1
1
)
und
er Cog
n
i
tiv
e Scien
c
e
Research
In
itiativ
e in
th
e Dep
a
rtm
e
n
t
o
f
Co
m
p
u
t
er Scie
n
ce, Am
rita Sch
o
o
l
of Arts an
d
Scien
ces,
Am
rita
Vish
wa
Vidy
a
p
eetham
Uni
v
e
r
sity
, K
o
chi
.
REFERE
NC
ES
[1]
Mao Lin Huan
g, Liang Fu Lu, Xu
y
un
Zhan
g. “Using ar
ced
axes in par
a
llel coord
i
nates geometr
y
for high
dimensional B
i
g
D
ata v
i
sual analy
t
ics
in
cloud
co
mputing”, Sprin
g
er-Verlag
Wien
, 2014
.
[2]
Hong Zhou, Xiaoru Yuan2, Huamin Qu, Weiwei Cui, B
a
oquan
Chen, “Visual C
l
ustering
in Parallel Coordinates
”
,
Eurographics/ I
EEE-
VGTC Sym
posium on Visua
lization,
Vol. 27, No. 3
,
2008
.
[3]
Michael Schroed
e
r, David Gilbert, Jac
ques van Helden, Penn
y
No
y
,
“Approach
es
to visualization
in bioinformatics
:
from dendrograms to Space Exp
l
orer”,
2000
.
[4]
Almir Olivette
Artero, Mar
i
a C
r
istina Ferr
eir
a
d
e
Olivei
ra, and
Haim Levkowitz, “Unc
overing C
l
usters in Crowd
e
d
Parall
el Coord
i
n
a
tes Visua
liz
at
io
ns”,
IEEE Symp
osium on Information
Visualization
, 2004
.
[5]
Ronald R
.
Yag
e
r
,
Dimitar
P. Filev, “Summarizing da
ta using
a similarity
ba
sed
mountain
method”, 2007
.
[6]
Hadley
Wickham Bin,
“A framework for visu
alis
ing large data”,
t
r
anstats,
2013.
[7]
Aritra Dasgupta, Min Chen,
and
Robert
Kosara,
“Conceptualizin
g Visual Uncer
ta
inty
in
Parallel
Coordinates”, The
Eurographics Association
and
Bl
ackwell Publishing Ltd
.
, 2012.
[8]
Hani Siir
tola
. “
D
irec
t Manipu
la
ti
on of Par
a
ll
el Co
ordinat
e
s”,
IEEE
, 2000
.
[9]
Jimmy
Johansson, Patr
ic Ljung, Mikael Jern,
Matthew Cooper, “Rev
ealing S
t
ructur
e within
Clustered
Par
a
llel
Coordinates Display
s
IEEE S
y
mpos
ium on Information Visualization”,
I
EEE
,
200
5.
[10]
Ra
y
n
er Alfred
,
“
D
is
cretiza
tion
Num
e
rical Dat
a
fo
r Relat
i
onal
Data with One-
to-Man
y
R
e
la
ti
ons“
,
Journal of
Computer Scien
c
e
, Vol. 5, No. 7
,
pp
. 519-528
, 2
009.
[11]
Helwig
Hauser
, Florian Ledermann,
an
d
Helmut
Doleisch, “Angular Brushing
of
Extend
ed Par
a
llel Coordin
a
tes”.
[12]
Inselberg
and B. Dimsdale, “Paralle
l Coordinates: A Tool For Visual
i
z
ing Mult
idi
m
e
nsi
onal Geometr
y
”,
IE
EE Con
f
.
on Vis. '90,
pp.3
61-378, 1990
.
[13]
Ying-Huey
Fua, Matthew O. Ward,
and
Elke A. Rundenstein
er,
“Hierarchical
Parallel Coordin
a
tes for Exploratio
n
of Larg
e Da
tas
e
t
s
”.
[14]
Aritra Dasgupta
and Rober
t
Kosara, “Screen-
Space Metr
ics for
Par
a
llel Coord
i
nates”.
[15]
M
y
ung-Hoe Hu
h, Dong Yong P
a
rk, “Enh
ancing
parallel
coord
i
nate p
l
ots”, Publis
hed b
y
Elsev
i
er
Ltd.
[16]
Pak Chung Wong, and
R. Daniel
Bergeron, “Multiresolution
Mul
tidimensional Wavelet Brushing”,
IEEE
, 1996
.
[17]
Hing Yan lee an
d Hwee-Leng O
ng, “Vis
ualizatio
n Support for
Data Mln
l
ng”,
IEEE
,
1996
.
[18]
Allen R
.
Mar
tin,
Matth
ew O. W
a
r
d, “
H
igh Dim
e
n
s
ional Brus
hing
for Inter
a
c
tive
E
xploration
of M
u
ltivar
ia
te Da
ta
”
,
IEEE, 1995
.
[19]
Tian Zh
ang, Raghu Ramakrishnan, Miron Livn
y
,
“BIRCH:
An Efficient Data Cluste
ring Meth
od for Ver
y
Lar
g
e
Da
ta
ba
se
s”,
SIG
M
OD ’96
, Montreal, C
a
nada IQ,
1996.
[20]
Matthew O. Ward, “Xm
dvTool:
Integrating Mult
iple
Met hods fo
r Visualizi
ng M
u
ltivar
iate Data”,
IE
EE,
pp. 1070
-
2385, 1994
.
[21]
J. Han, and
M.
Kamber, “Data
Mining: Con
cepts a
nd Techn
i
ques”, Morgan
Kauf
mann, San Fran
cisco, 2001
.
[22]
Sturges H. A., "
T
he
choice
of
a
class in
terva
l
".
Journal of
the Am
erican S
t
atistical Association,
pp. 65–66, 1926.
Evaluation Warning : The document was created with Spire.PDF for Python.