Internati
o
nal
Journal of Ele
c
trical
and Computer
Engineering
(IJE
CE)
V
o
l.
6, N
o
. 4
,
A
ugu
st
2016
, pp
. 18
18
~
1
827
I
S
SN
: 208
8-8
7
0
8
,
D
O
I
:
10.115
91
/ij
ece.v6
i
4.1
029
5
1
818
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
Segm
entation of
Ret
a
il M
o
bile M
a
rket
Usi
n
g HMS Algorith
m
Koyi Anush
a
,
Yas
h
aswini C,
Manish
ank
a
r
S
Department o
f
C
o
mputer Scien
c
e, Amrita Sc
hool
of Arts and
scien
ces, M
y
suru C
a
mpus,
Amrita Vishwa
Vidy
apeetham
,
Am
rita Universit
y
,
Indi
a
Article Info
A
B
STRAC
T
Article histo
r
y:
Received Feb 25, 2016
Rev
i
sed
May 21
, 20
16
Accepte
d
J
u
n 9, 2016
In the modern
world of marketing, an
al
yz
ing t
h
e trends in m
a
rket is a
k
e
y
point towards to scope of improvement
of an
y compan
y
.
Considering the
anal
ys
is
of a r
e
t
a
il m
a
rke
t
is
hig
h
l
y
chal
leng
ing where m
a
rket tr
ends
chang
e
ver
y
fr
equen
t
ly
based on custo
m
er n
e
e
d
s a
nd inte
re
st.
Ma
rke
t
se
gme
n
ta
tion
is one of the approaches included
in an
aly
s
is of market tr
ends which gives a
divers
e v
i
ew of
the m
a
rke
t
.
The
res
ear
ch h
e
re
c
oncentr
ates
, es
p
eci
all
y
on
a
case stud
y
b
a
sed on fast movin
g
consum
able g
oods market an
d identif
y
ing
market ch
ange
patterns b
y
app
l
y
i
ng
a novel d
a
ta mining
appr
oach. Data
mining includ
es a wide variety
of t
echn
i
ques and algorithm which can be
effec
tive
l
y
us
ed
in the proces
s
of m
a
rket anal
y
s
is
. The res
ear
ch work carrie
d
out coins
a new algorithm wh
ich combin
es various association rules an
d
techn
i
ques
,
the HM
S
(Hy
b
rid M
a
rket
Segm
en
tation)
algor
ith
m
with som
e
specialized
criteria
is used to
su
ppor
t th
e market segmentation.
The pr
imar
y
data need
ed fo
r the
analy
s
is and op
eration
are collected
through
a
questionnaire based survey
cond
ucte
d on p
e
ople from various demographic
regions as well as various age gr
oups.
Used a quota based sampling approach
for the research
, Th
e d
a
ta mining approach h
e
re helps to stud
y
the larg
e
datas
e
t col
l
ec
ted
and als
o
to ex
tr
act th
e useful
inf
o
rmation requir
e
d to model
the s
y
s
t
em
. The
s
y
s
t
em
here is
a
learn
i
ng s
y
s
t
em
which im
proves
the m
a
rket
segm
entation
fu
nction
a
lit
y
as d
a
ta
set
im
proves, The
paper
i
m
p
lem
e
nts a
h
y
brid
data min
i
ng appro
ach w
h
ich
e
ffec
tiv
el
y segm
ents the
r
e
ta
il m
obil
e
market in
to v
a
rious customer
a
nd produ
ct g
r
oups and also provides
a
prediction
and s
uggestion s
y
s
t
em for compan
y
as well
as custom
er.
Keyword:
Clu
s
tering
Dem
ogra
phi
c s
e
gm
ent
a
t
i
on
H
M
S algor
ithm
Mark
et seg
m
e
n
tatio
n
Copyright ©
201
6 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
:
Koy
i
An
us
ha,
Depa
rt
m
e
nt
of
C
o
m
put
er Sci
e
nce,
Am
rita Vishwa
Vidy
a
p
eetham
U
n
ive
r
sity
, M
y
sur
u
Cam
pus,
#
114
,7
th
C
r
oss,
B
o
gadi
2
nd
Stag
e, Mysu
ru
-570
026
.
Em
a
il: ch
o
w
d
a
ryano
o
s
h
a
@gmail.co
m
1.
INTRODUCTION
In a
n
y
ret
a
i
l
m
a
rket
cust
om
er i
s
ki
n
g
, u
n
d
erst
a
ndi
ng c
u
st
om
er needs
and i
n
t
e
rest
s i
s
t
h
e
m
o
st
ch
allen
g
i
ng
task
in th
e field
of m
a
rk
etin
g.
Mark
et an
al
ysis m
a
in
ly co
n
cen
trates
o
n
areas lik
e id
en
tifyin
g
the
pu
rc
hase
pat
t
e
rn
o
f
a
c
u
st
om
er, e
x
t
r
act
i
n
g
cust
om
er p
r
o
f
i
l
e an
d s
u
ggest
i
n
g
p
r
o
d
u
ct
s
f
o
r
cu
st
om
ers,
fi
n
d
i
n
g
th
e pro
d
u
c
t
de
m
a
n
d
v
e
rsu
s
sales ratio.
Retail
m
a
rk
eti
n
g co
m
p
rises th
e selling
go
od
s
and
services to
co
nsu
m
er
s th
ro
ugh
d
i
str
i
bu
ted
ch
ann
e
ls t
o
ear
n
m
o
r
e
pr
ofit. Retail
m
a
rk
et v
a
ries a l
o
t fro
m
wh
o
l
esale i
n
it is
ope
rat
i
o
nal
be
havi
or
, ret
a
i
l
m
a
rket
co
nce
n
t
r
at
es o
n
num
ero
u
s
p
r
o
d
u
ct
s sol
d
o
u
t
i
n
t
o
vari
ous
cu
st
o
m
ers i
n
sm
a
ller q
u
a
n
tities, wh
ereas a
who
l
esale m
a
r
k
et p
r
ov
id
es a
b
u
l
k
produ
ct sto
c
k
sale to
b
i
g
g
e
r org
a
n
i
zatio
ns
or
industries.
The
r
e are
differe
n
t form
s
of
ret
a
i
l
i
ng suc
h
as
sh
op
base
d,
e-
comm
erce, direct
m
a
rketing, etc.
ecommerce concent
r
ates on
marketing a
n
d sales, whic
h takes place
through (B2C
) business
-to-c
o
nsum
e
r
t
r
ansact
i
o
ns i
n
an o
n
l
i
n
e s
h
op
pi
n
g
po
rt
al
or t
h
r
o
u
g
h
m
a
i
l
orde
rs. R
e
t
a
i
l
cust
om
ers are m
o
re dem
a
ndi
ng
expecte
d
suc
h
as availabilit
y, value, c
hoi
ces, acce
ssibility and pricing etc., in order to ful
f
ill t
h
ese
requ
irem
en
ts o
f
th
e cu
sto
m
er’s retailer is fo
rced
to
c
h
a
nge t
h
ei
r l
a
n
d
s
cape ve
ry
ra
pi
dl
y
.
To sat
i
s
fy
t
h
e
Evaluation Warning : The document was created with Spire.PDF for Python.
I
J
ECE
I
S
SN
:
208
8-8
7
0
8
Seg
m
e
n
t
a
t
i
o
n
of
Ret
a
i
l
M
o
bi
l
e
Market
Usi
n
g
HMS
Al
g
o
ri
t
h
m
(
K
oyi
A
nus
ha)
1
819
custom
er re
qui
rem
e
nts the re
tailers
are
planning
for t
h
e
new st
rategies s
u
ch as
ne
w
business m
odels, ne
w
com
m
uni
cat
i
on, ne
w ope
rat
i
o
n
et
c.
Market segm
entation is the process of categorizing
t
h
e m
a
rket
i
n
t
o
di
ffe
rent
cust
om
er gr
o
ups a
n
d
p
r
od
u
c
t
g
r
oups,
b
a
sed
o
n
sev
e
r
a
l
pr
ef
er
ences su
ch as, ed
u
cation
,
in
come, r
e
g
i
o
n
, cou
n
t
r
y
, etc.
wh
ich
ar
e
depe
n
d
ent
a
n
d
i
ndepe
n
d
ent
param
e
t
e
rs aff
ect
i
ng t
h
e cha
nge
s i
n
t
h
e m
a
rket
, m
a
rket
segm
ent
a
t
i
on hel
p
s i
n
un
de
rst
a
n
d
i
n
g
t
h
e cust
om
er needs a
nd i
n
t
e
r
e
st
s of a
n
i
ndi
vi
d
u
al
ve
ry
ef
f
i
ci
ent
l
y
for t
h
e
deci
si
o
n
m
a
ki
ng i
n
the
m
a
rket, and also providi
ng a
n
ex
act p
i
ctu
r
e on
bu
ying
p
a
ttern
s o
f
a
custom
er. Ma
rket segm
entation is
cl
assi
fi
ed i
n
t
o
vari
ous t
y
pe
segm
ent
a
t
i
ons l
i
k
e dem
o
g
r
ap
hi
c segm
ent
a
t
i
on,
Ge
og
rap
h
i
c
segm
ent
a
t
i
o
n
,
B
e
havi
oral
se
g
m
ent
a
t
i
on,
Psy
c
ho
g
r
ap
hi
c se
g
m
ent
a
t
i
on,
occ
a
si
onal
se
gm
ent
a
t
i
on, c
u
l
t
u
ral
segm
ent
a
t
i
on [
1
]
.
I
d
en
tif
yin
g
th
e up
-g
row
i
ng
tren
ds in
t
h
e
f
i
eld
of
m
a
rk
eting
is
requ
ired
i
n
th
e bu
sin
e
ss p
r
o
cess, as
wel
l
as Ext
r
ac
t
i
ng k
n
o
w
l
e
dg
e, i
n
f
o
rm
at
i
on fr
om
a huge cust
om
er dat
a
set
i
s
very
im
port
a
nt
i
n
b
u
si
ne
ss f
o
r
decision m
a
king a
n
d
busi
n
ess
proce
ss.
Sales
pattern i
n
the
inve
ntory and
forecasting the
great
pote
n
tial suc
h
as d
ecision
m
a
k
i
ng
, m
a
rk
et co
m
p
etitio
n
an
d strateg
i
c p
l
ann
i
ng
is an
essen
tial facto
r
in
t
h
e bu
sin
e
ss pro
cess
[2]
.
Dem
ogra
p
hi
c st
o
r
e w
h
i
c
h m
a
i
n
l
y
concentrates on
a particular
se
gment
of
th
e
g
l
obe and
pr
odu
cts th
at is
part
i
c
ul
a
r
t
o
a
dem
ograp
hi
c
area. Ge
ne
ral
st
ore- a
rural
sto
r
e th
at su
pplies n
eeds in t
h
e local community.
Su
perm
arket
i
s
a sel
f-se
r
vi
ce st
ore
w
h
i
c
h p
r
ovi
des al
l
vari
et
i
e
s of
a pr
o
duct
u
nde
r a si
ngl
e
ro
of
.
Hyp
e
rm
ark
e
ts-p
rov
i
d
e
s a
wide co
llectio
n of produ
cts wit
h
lo
w m
a
rg
in
s,
th
e cost is com
p
arativ
ely le
ss th
at
ot
he
r ret
a
i
l
fo
r
m
at
[3]
.
E-Tai
l
or c
u
st
om
er sho
p
a
nd
b
u
y
s
t
h
e p
r
o
d
u
ct
he
or s
h
e
want
s
t
h
r
o
u
g
h
t
h
e i
n
t
e
rnet
.
There
are
seve
ral factors
which s
h
oul
d
be re
m
e
m
b
ered
by
th
e
retailer to in
crease th
eir
sales and
few
o
f
th
e
im
port
a
nt
f
act
ors as
U
n
et
hi
c
a
l
C
ons
um
er Behavi
or i
s
t
h
e
f
act
or
whe
r
e i
n
t
h
e cust
om
er buy
s t
h
e
pr
od
uc
t
an
d
th
en
retu
rn
s it
b
ack
after t
h
e p
r
od
u
c
t fu
lfi
lled
its p
u
rpo
s
e. Cu
sto
m
er serv
ice is th
e
facto
r
wh
ich
help
s to
unde
rstand the
needs
of the c
u
stom
er and the retailer
as to
act accordi
ng t
o
it and fulfill all the require
ments
of t
h
e c
u
st
om
er an
d anal
y
z
e ho
w t
o
i
m
pro
v
e cu
st
om
er sat
i
s
fact
i
on [
4
]
.
In
de
pe
nde
nt
R
e
t
a
i
l
C
u
stom
er
Services- to a
n
alyze how t
o
extend a sm
all retailer
and
t
o
p
r
ovi
de a
go
od
ser
v
i
ce. M
a
l
e
/
F
em
al
e shop
pi
n
g
beha
vi
o
r-t
o a
n
al
y
ze ho
w
di
f
f
e
rent
l
y
, m
a
l
e
and
fem
a
l
e
sh
op
pi
n
g
be
havi
or
va
ri
es
Ho
w
t
o
cat
er
t
h
e
m
a
rket
,
according to t
h
ese categories
and t
h
us fi
ndi
ng the
de
pend
e
n
t and i
nde
pe
ndent va
riables in researc
h
on m
a
rket
segm
entation. Data
m
i
ning places a vital role in the an
alysis as well as
segm
entation process as it ta
kes all
these fact
ors
a
b
ove m
e
ntione
d as
input
t
o
f
o
rm
an infe
renc
e en
gine
[
5
]
.
Data m
i
n
i
n
g
is th
e
p
r
o
cess
o
f
ex
tracting
data fro
m
larg
e set o
f
raw
d
a
ta an
d
su
mm
arizin
g
it in
to
usef
ul in
fo
rm
ation f
o
r t
h
e f
u
rther
use
[6]
.
The a
n
aly
s
is step
of
KD
D
(K
now
ledg
e D
i
scov
er
y in
D
a
tab
a
ses)
p
r
o
cess is data min
i
n
g
. It is
an
an
alysis of
ex
trac
ting
large qu
an
tities of d
a
ta
fro
m
a prev
i
o
u
s
ly
u
nkno
wn
,
i
n
t
e
rest
i
ng
pat
t
erns s
u
c
h
as
A
ssoci
at
i
on R
u
l
e
M
i
ni
ng
(d
e
p
ende
nci
e
s)
, C
l
ust
e
ri
n
g
(g
ro
u
p
s o
f
dat
a
rec
o
rds
)
an
d
anom
al
y
det
ect
i
on
(u
n
u
sual
reco
rd
s).
Dat
a
m
i
ni
ng i
n
v
o
l
v
es si
x Common tasks: class
i
fication, cl
ust
e
ring,
reg
r
essi
o
n
, ass
o
ci
at
i
on r
u
l
e
l
earni
ng
, an
d a
nom
al
y
det
ect
ion
.
C
l
assi
fi
cat
i
on i
s
t
h
e t
a
sk
of
gr
ou
pi
n
g
k
n
o
w
n
st
ruct
u
r
e t
o
ap
pl
y
t
o
new
da
t
a
. C
l
ust
e
ri
ng
-
i
t
i
s
t
h
e t
a
sk of
gr
ou
pi
n
g
si
m
i
l
a
r st
ruct
ure
s
t
o
a si
ngl
e gr
o
u
p
with
ou
t
u
s
ing
k
nown stru
ctures in
d
a
ta. Reg
r
essi
on
-it is
t
h
e task
o
f
attem
p
t
i
n
g
to fi
n
d
a fun
c
tio
n in
wh
ich
d
a
ta with
less erro
r.
Asso
ci
atio
n
ru
l
e
l
ear
ni
n
g
-i
t
i
s
al
so
kn
o
w
n as
de
pe
nd
en
cy m
o
d
e
lin
g
,
it is th
e task
of
search
i
n
g th
e relatio
n
s
h
i
p
b
e
tween th
e v
a
riab
les [7
].
A
n
o
m
aly d
e
tectio
n
-
it
is also kno
wn as
o
u
tlier/d
e
v
i
atio
n
det
ect
i
on/
c
h
an
ge. It
i
s
t
h
e t
a
sk o
f
i
d
e
n
t
i
f
y
i
ng
u
nus
ual
da
t
a
sets, th
at
m
a
y b
e
in
teresting
or d
a
ta errors th
at
requ
ired
to fu
rt
h
e
r u
s
e.
B
r
i
t
o
, Pe
dr
o
Quel
has, et
.al
.
“C
ust
o
m
e
r segm
ent
a
t
i
on i
n
a l
a
rge
dat
a
ba
se of a
n
onl
i
n
e cust
om
i
zed
fashi
on
bu
si
ne
ss” t
h
i
s
pape
r
pr
ovi
des cus
t
om
er seg
m
ent
a
t
i
on i
n
fas
h
i
on b
u
si
ness
base
d o
n
cust
om
er
pre
f
ere
n
ces l
i
ke a
g
e,
gen
d
e
r
et
c. T
h
ey
h
a
d
used
t
w
o
di
ffe
re
nt
t
ech
n
i
ques t
o
gr
ou
p cat
eg
ori
e
s
o
f
sam
e
cust
om
ers.
On
e i
s
cl
ust
e
ri
ng
and
sec
o
n
d
o
n
e
i
s
su
b
gr
o
u
p
di
sc
ove
ry
. B
a
sed
on
t
h
ese
m
e
t
hods
o
r
ga
n
i
zat
i
on
can
easily under
s
tand
w
h
at ty
p
e
o
f
pr
odu
cts
cu
sto
m
er
pr
ef
er
s an
d b
a
sed
on
th
at
o
r
g
a
n
i
zatio
n
can
m
a
n
u
f
actu
r
e
the products [8]. Aditya J
o
shi
e
t et al.
“use
o
f
dat
a
m
i
ni
ng t
e
chni
que
s t
o
i
m
pr
o
v
e t
h
e e
f
f
e
c
t
i
v
eness
o
f
sal
e
s an
d
m
a
rket
i
ng” T
h
i
s
paper
pr
o
p
o
s
es cl
ust
e
r ass
o
ci
at
i
on m
i
ni
ng ap
pr
oac
h
t
o
cl
assi
fy
a
ssociated patterns
of sale
and cl
assi
fy
st
ock
dat
a
. To i
m
pl
em
ent
t
h
e
beha
vi
o
r
t
h
ey
uses t
w
o pha
se
, fi
rst
phas
e
di
vi
de t
h
e st
oc
k i
n
t
h
re
e
d
i
fferen
t
clu
s
t
e
rs
o
n
th
e
b
a
sis o
f
so
ld
quan
tities su
ch as DS(d
ead stock
)
, SM
(slow-m
o
v
i
ng
), FM(fast-
m
ovi
ng)
.i
n t
h
e
seco
nd
p
h
ase
t
h
ey
pr
o
p
o
s
ed
M
FP(m
o
st
fre
que
nt
pat
t
e
rn
)a
l
g
o
r
i
t
h
m
whi
c
h fi
nd
s f
r
e
que
n
c
i
e
s of
p
a
ttern v
a
l
u
es of the co
rresp
ond
in
g
items [9
]. Migu
ei
s et al. “cu
s
t
o
m
e
r d
a
ta m
i
n
i
ng
for lifestyle
segm
ent
a
t
i
on” Thi
s
pa
per
pr
ovi
des l
i
f
e
st
y
l
e segm
ent
a
t
i
on
base
d
on
l
i
f
e st
y
l
e of
t
h
e
cust
om
ers. T
h
ey
had
u
s
ed
VARCLUS al
g
o
rith
m
to
cl
u
s
ter t
h
e sa
m
e
g
r
o
up of
p
e
rson
s. Still th
ey
h
a
d u
s
ed
h
i
erarch
ical al
g
o
rith
m
whi
c
h com
b
i
n
es di
vi
si
bl
e and a
ggl
om
erati
v
e al
go
ri
t
h
m
s
for cl
u
s
t
e
ri
n
g
[10]
.
No
ori
,
B
e
hr
ooz et
al
.
“A
n
anal
y
s
i
s
of m
obi
l
e
ban
k
i
n
g
u
s
er be
ha
vi
o
r
u
s
i
ng c
u
st
om
er segm
ent
a
t
i
on” t
a
ki
ng a ca
se
st
udy
o
f
a
n
I
r
a
ni
an
ban
k
t
o
fi
n
d
t
h
e cust
om
er usa
g
e
of
t
h
e
o
n
l
i
n
e ba
n
k
i
n
g.
Usi
n
g
a m
e
t
hod
ol
ogy
o
f
R
F
M
D
t
o
fi
nd
t
h
e
rec
e
ncy
,
f
r
e
q
u
e
n
c
y
o
f
t
h
e custo
m
er
s an
d
t
o
id
en
tif
y p
u
r
c
h
a
sed of
d
i
f
f
e
r
e
n
t
pr
oducts th
r
oug
h onlin
e m
o
b
ile b
a
n
k
i
n
g
[1
1]
. B
a
ra
dwa
j
, B
.
K, et
al
. “M
i
n
i
ng E
d
u
cat
i
onal
Dat
a
t
o
A
n
al
y
ze St
ude
nt
s Pe
rf
or
m
a
nce”
in
the p
a
p
e
r
di
scuss
e
d c
u
r
r
e
nt
si
t
u
at
i
on a
s
wel
l
as fut
u
r
e
. They
ha
d
us
ed di
f
f
e
r
ent
d
a
t
a
m
i
ni
ng t
ech
ni
q
u
es t
o
pre
d
i
c
t
t
h
e
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
SN
:
2
088
-87
08
I
J
ECE
Vo
l. 6
,
N
o
. 4
,
Au
gu
st 2
016
:
18
18
–
1
827
1
820
stu
d
e
n
t
p
e
rfo
r
man
ce in
all a
c
tiv
ities su
ch
as ex
am
s an
d
g
a
m
e
s etc. Ma
in
ly u
s
ed
ID3d
ecision
tree to
find
a
best
p
o
ssi
bl
e
way
t
o
o
r
ga
ni
z
e
a l
earni
n
g
se
t
i
n
quest
i
o
ns
.
Dat
a
ha
d col
l
e
ct
ed fr
om
one
uni
versi
t
y
an
d
st
ore
d
i
n
f
o
rm
at
i
on.
B
a
sed o
n
pre
v
i
o
us dat
a
sy
st
em
pre
d
i
c
t
s
t
h
e st
u
d
ent
s
’ per
f
o
r
m
a
nce [1
2]
.
Liu
,
Xing
li, an
d
Hu
ali Liu
.
“An
Im
p
r
ov
ed
Ap
riori Algo
rith
m
fo
r Asso
ciatio
n
Ru
les” th
is p
a
p
e
r
expl
ai
n
s
a
b
o
u
t
t
h
e i
m
pro
v
ed
apri
ori
al
go
ri
t
h
m
i
n
associ
at
i
o
n
rul
e
m
i
ni
ng,
t
h
i
s
i
m
prove
d
apri
ori
al
go
ri
t
h
m
as
im
proved the
efficiency and it esta
bl
i
s
hes a new dat
a
ba
se t
o
sim
u
l
a
t
e
appl
i
e
d ex
per
i
m
e
nt
consi
s
t
i
ng
of
stu
d
e
n
t
ach
i
eve
m
en
t in
co
m
p
u
t
er
p
r
og
rammin
g
cou
r
se [1
3
]
. Zhou
,
Qish
en et al. “In
t
ellig
en
t Data Mi
n
i
ng
and
Decision System for Comm
ercial Deci
sio
n
Mak
i
ng
” th
is pap
e
r tells th
e im
p
o
r
tan
ce of th
e inform
at
io
n syste
m
and t
h
e data
ba
se, which is
ve
ry
m
o
st im
port
a
nt in th
e
business for the
de
cision m
a
king
according to t
h
e rapi
d
ch
ang
e
s
b
een
co
nsid
ered
in
th
e bu
sin
e
ss, it co
n
s
id
ers th
e
m
o
st fiv
e
th
emes to
u
n
d
e
rst
a
n
d
t
h
e co
m
p
etitiv
e
change in t
h
e
busi
n
ess strate
gies suc
h
as le
vera
ge th
e
fra
nchise
, intensi
f
y non-a
p
parel, accelerate retail- led
gr
owt
h
, i
nvest
i
n
u
nde
r-
pe
n
e
t
r
at
ed m
a
rket
and
p
u
rs
ue
o
p
erat
i
o
nal
exc
e
l
l
e
nce [1
4]
.
Hipp, J
o
chen, Ulrich
Guntzer et.al.
“Algorithm
s
for ass
o
ci
at
i
o
n
r
u
l
e
m
i
ni
ng-a
g
e
neral
s
u
rvey
and
com
p
arison” pape
r provides
t
h
e
com
p
ari
s
on
st
udy
bet
w
ee
n
d
i
ffere
nt
al
g
o
ri
t
h
m
s
used i
n
market analysis. Use
d
a
n
algorithm
,
and
analyzed
t
h
ei
r
per
f
o
r
m
a
nce
base
d o
n
r
unt
i
m
e experi
m
e
nt
s wi
t
h
m
a
i
n
t
a
i
n
i
n
g
t
h
res
hol
d
val
u
e.
Pa
per c
o
ncl
u
de
d
t
h
at
i
n
mark
et an
alysi
s
asso
ciation
ru
le work
s b
e
tter th
an
o
t
h
e
r al
go
rith
m
s
[15
]
.
2.
R
E
SEARC
H M
ETHOD
Un
de
rst
a
n
d
i
n
g
t
h
e cust
om
er
sat
i
s
fact
i
on rat
i
o and
b
u
y
ing p
a
ttern
is th
e
m
a
j
o
r con
c
ern
of all th
e
maj
o
r
retail mark
eting
co
m
p
an
ies. Mark
et
seg
m
en
tatio
n
an
d an
alysis plays a v
ital ro
l
e
in
th
is con
t
ex
t, th
e
exi
s
t
i
ng t
ech
ni
que
s i
n
t
h
ese fi
el
ds di
d
not
gi
ve ade
q
u
a
t
e
care to analyze the interrelat
ed param
e
ters that plays
a
m
a
jor r
o
l
e
i
n
que
ui
n
g
a sol
u
t
i
on t
o
t
h
e m
a
rket
segm
ent
a
t
i
on
. The
r
e ha
s been a
para
di
g
m
shi
f
t
t
h
at
has t
a
ke
n
place in the m
a
rket, which gi
ves t
h
e
ou
tlook
only to the
retail sector a
n
d
alternating c
u
stom
er be
havior. The
maj
o
r ch
allen
g
e th
at lies
in
t
h
is is co
in
in
g
an
alg
o
r
ithm
whic
h mixes all aspects of a
m
a
rket factors as an
in
pu
t and
g
e
n
e
rates an ou
tpu
t
for
p
r
ed
ictio
n
o
r
sugg
estion
of m
a
rk
et related
d
e
v
i
ation
s
.
Th
e in
itial step
th
at h
a
s b
e
en
tak
e
n
t
o
wards th
e sy
stem framin
g
was creating
a u
s
er in
terface for bo
t
h
the custom
er and c
o
m
p
any to record
th
eir
mark
et related tran
sactio
ns
. These
i
n
put
s were pre
p
r
o
ce
ssed
an
d
sto
r
ed
in
t
o
a
d
a
tab
a
se fo
r th
e fu
t
u
re
u
s
age o
f
th
e
i
n
feren
ce en
g
i
n
e
.
Th
e infere
n
ce en
g
i
n
e
fun
c
tionality is
bi
f
u
rcat
ed i
n
t
o
t
w
o
di
st
i
n
ct
b
u
t
i
n
t
e
rc
on
nect
ed m
odul
es t
h
e su
ggest
i
o
n a
nd
p
r
edi
c
t
i
o
n s
y
st
em
as wel
l
as t
h
e
analysis engine
.
Pre
d
iction a
n
d suggestion sy
ste
m
works on the basis
of the p
r
ev
iou
s
ly an
alyzed
tran
sactio
n
d
e
tails,
whi
c
h
were
p
r
ovi
ded
by
t
h
e
dat
a
col
l
ect
o
r
.
Thi
s
part
has s
o
m
e
i
n
t
e
gral
al
go
ri
t
h
m
s
pr
ofi
c
i
e
nt
i
n
han
d
l
i
n
g
t
h
e
i
n
t
e
rrel
a
t
e
d
dat
a
fr
om
cust
om
er a
n
d
com
p
an
y
,
ge
ne
rating
a p
a
ttern
for easier pred
iction
an
d sugg
estion
.
The arc
h
itecture diagram
de
note
d
as
Fig
u
re 1
ai
m
s
to
p
o
r
tray th
e in
terconn
ectio
n
that
m
a
k
e
s th
e
un
de
rl
y
i
ng pl
a
t
form
for t
h
e
pr
o
pose
d
sy
st
em
. It
al
so sho
w
s t
h
e va
ri
o
u
s
com
ponent
s a
nd t
h
e
dat
a
fl
o
w
t
h
at
happe
n
s i
n
side
the system
.
Figure 1.
Archi
t
ecture diagra
m
Evaluation Warning : The document was created with Spire.PDF for Python.
I
J
ECE
I
S
SN
:
208
8-8
7
0
8
Seg
m
e
n
t
a
t
i
o
n
of
Ret
a
i
l
M
o
bi
l
e
Market
Usi
n
g
HMS
Al
g
o
ri
t
h
m
(
K
oyi
A
nus
ha)
1
821
Anal
y
s
i
s
e
ngi
n
e
has a
key
r
o
l
e
as i
t
per
f
o
r
m
s
t
h
e se
gm
ent
a
t
i
on
part
one
o
f
t
h
e p
r
om
i
n
ent
goal
s
o
f
t
h
e
pr
o
pose
d
sy
st
e
m
. M
a
rket
segm
ent
a
t
i
on t
o
ol
i
s
used t
o
di
ff
erent
i
a
t
e
t
h
e
v
a
ri
o
u
s cat
eg
ori
e
s of c
u
st
om
ers t
h
at
p
r
ev
ail in
t
h
e
retail
m
a
rk
et.
The a
d
m
i
ni
st
rat
o
r i
s
res
p
onsi
b
l
e
f
o
r t
h
e i
n
f
o
rm
at
i
on rega
r
d
i
n
g t
h
e
pre
v
i
ous t
r
ans
act
i
o
n
t
a
ken
pl
ace
with the sam
e
kind of products wh
ich the
new cust
om
er i
s
willing to buy,
according to the param
e
ters tha
t
have
bee
n
co
n
s
i
d
ere
d
i
n
t
h
i
s
wo
rk
.
W
e
per
f
o
rm
t
h
e pre-
pr
ocessi
n
g
t
ech
n
i
ques
on t
h
e t
o
p o
f
t
h
e dat
a
i
n
o
r
de
r
to
rem
o
v
e
redun
d
a
n
t
, con
s
isten
c
y and
nu
ll d
a
ta in
th
e
co
llected
d
a
ta to
g
e
t ex
act d
a
ta.
The ne
xt step i
n
the process is to analyze the data
with
rel
a
ted
in
fo
rm
atio
n
alread
y
g
a
th
ered
fro
m
th
e
adm
i
n, per
f
o
r
m
di
fferent
al
g
o
ri
t
h
m
s
such
a
s
ap
ri
ori
al
go
ri
t
h
m
and cl
ust
e
ri
n
g
t
ech
ni
q
u
es
ha
ve
been
use
d
t
o
get
com
m
on usage
of
p
r
o
d
u
ct
s by
t
h
e cust
om
er. B
a
sed o
n
t
h
e an
alyzed
informatio
n
it is easy to
g
e
t seg
m
en
tatio
n
o
f
th
e cu
st
o
m
e
r
b
e
h
a
v
i
or t
h
ro
ugh
u
s
ing d
i
fferen
t
p
a
ra
m
e
t
e
rs.
At t
h
e end
seg
m
en
ted
an
alysis will h
a
v
e
t
h
e
tran
saction
info
rm
atio
n
o
f
prev
iou
s
cu
sto
m
ers, its d
i
r
ectly
sto
r
ed
informatio
n
to
t
h
e storage se
rver.
For ne
w
user
A
r
c
h
i
t
ect
ure
hel
p
s t
o
i
d
ent
i
f
y
w
h
at
t
y
pe
of
p
r
od
uct
s
ha
ve t
o
b
u
y
and
i
t
se
rves
the
requirem
ents and
sat
i
s
fy
i
ng t
h
e
cust
om
er by
s
e
rvi
ng t
h
em
wi
t
h
best
qual
i
t
y
of se
r
v
i
ces a
n
d
p
r
o
d
u
ct
s. T
h
e sy
st
em
pro
v
i
d
es a
platform
for the c
o
m
p
any to int
r
oduce their ne
w
prod
ucts with t
h
e
m
odified quality according t
o
t
h
e
custom
er prefe
r
ences
and like
s
.
The p
r
o
p
o
se
d sy
st
em
i
s
been
desi
g
n
ed
by
usi
n
g dat
a
m
i
ni
n
g
t
echni
q
u
e
such as A
sso
ci
at
i
on R
u
l
e
Min
i
n
g
wh
ich
h
e
lp
s t
o
id
en
tify th
e in
terest pattern
s b
e
t
w
ee
n t
h
e
vari
a
b
l
e
s i
n
a h
uge
dat
a
base, t
h
e al
g
o
r
i
t
h
m
wh
ich
is
b
e
en
u
s
ed
is t
h
e
HMS algo
rith
m
wh
ich
is
h
e
l
p
s
min
i
n
g
t
h
e
freq
u
e
n
t
item
set, th
e al
g
o
rith
m
h
e
lp
s to
bri
ng
out
t
h
e l
i
s
t
of i
t
e
m
s
been p
u
rc
hase
d b
y
t
h
e ol
d cust
o
m
er and hel
p
s
t
o
recom
m
end t
h
e new c
u
st
o
m
er t
o
kn
o
w
w
h
at
ki
n
d
o
f
pr
o
duct
s
a
r
e bee
n
p
u
rc
ha
sed,
whi
c
h hel
p
s t
h
e ne
w cu
s
t
om
er t
o
deci
de abo
u
t
t
h
e pr
o
duct
s
buy
i
n
g. T
h
e ot
her t
ech
ni
q
u
e
whi
c
h i
s
used
i
s
segm
ent
a
t
i
o
n, Se
gm
ent
a
t
i
on i
s
t
h
e pr
oces
s whe
r
e i
t
hel
p
s t
h
e
m
a
rket
t
o
di
vi
de i
n
t
o
a ce
rt
ai
n g
r
o
u
p
o
r
a
n
i
ndi
vi
d
u
al
, t
h
i
s
di
vi
si
o
n
of c
u
s
t
om
ers hel
p
s t
h
e ret
a
i
l
e
rs i
m
pr
o
v
e
th
eir bu
siness
an
d
fo
cu
s
on
th
e sales of the ite
m
in
a p
r
o
f
itab
l
e
way.
It also
h
e
lp
s i
n
und
erstan
d
i
ng
th
e
indivi
dual
or a
group of sim
ilar likin
g c
u
stomers and serve
s
them
accordi
n
g
to thei
r re
quirem
ents and
needs.
There
are
m
a
ny categorizing techniqu
es in
wh
ich
we can
seg
m
en
t th
e cu
st
o
m
ers, in
th
is
wo
rk
we h
a
v
e
con
s
i
d
ere
d
fe
w of t
h
e pa
ra
m
e
t
e
rs t
o
segm
ent
t
h
e cust
om
er and ful
f
i
l
l
t
h
ei
r requi
re
m
e
nt
s and nee
d
s, t
h
e
param
e
t
e
rs whi
c
h
we co
n
s
i
d
ere
d
are a
g
e,
gen
d
er
,
occu
pat
i
o
n, i
n
com
e
, educa
t
i
on,
beha
vi
o
r
t
h
ese
segm
ent
a
t
i
on h
e
l
p
ou
r w
o
r
k
i
n
segm
ent
i
ng
what
t
y
pe o
f
cust
om
ers buy
what
ki
nd
of
m
obi
l
e
phon
es
and h
o
w
to
in
crease t
h
e
sales of th
e prod
u
c
t and
also
help
s in m
o
d
e
lin
g a
n
e
w m
o
b
ile p
hon
e.
2.
1.
Hybrid Market Se
g
mentati
on Algorithm
H
M
S algor
ithm
(
∑
D,
α
,
β
,
π
,
Ω
)
∑
D
–
I
s th
e inpu
t d
a
ta
set
α
-i
s t
h
e s
u
ppl
e
m
ent
i
ng
param
e
t
e
r 1
β
-is the
su
p
p
o
r
ting
param
e
ter fo
r
α
π
-is th
e sim
ilar
ity p
a
ram
e
ter fo
r
α
and
β
fo
r th
e
d
a
ta inp
u
t
∑
D
Ω
-i
s t
h
e
rat
i
o
of
di
s
p
ari
t
y
val
u
e am
ong
α
and
β
Th
e algo
rith
m
tak
e
s an
inpu
t d
a
ta set co
llected
, and
ap
p
l
i
e
s th
e fo
ur p
a
ra
m
e
ters
α
,
β
,
π
, and
Ω
in
vary
i
n
g m
a
nn
er acc
or
di
n
g
t
o
t
h
e
r
u
l
e
f
o
r
segm
ent
a
t
i
o
n
l
i
k
e
dem
ogra
p
hi
c,
Geo
g
r
ap
hi
c, B
e
hav
i
oral
,
Psychographic, occasi
onal, cultural segm
entation a
n
d
fram
e the pa
ram
e
ter scale for eac
h
of the
scena
r
io.
Th
e
p
r
o
cedur
e is as f
o
llow
s
b
y
tak
i
ng
input an
d
m
a
tch
i
n
g
it w
ith
th
e t
e
st case seg
m
e
n
ts and
th
en
pr
o
v
i
d
i
n
g a
sc
ore
val
u
e i
n
e
ach case
.
T
h
e
fi
nal
sc
ore
val
u
e
whi
c
h i
s
an
ag
gre
g
at
e
of
t
h
e i
ndi
vi
d
u
al
s
c
ores
h
e
lp
s in
creatin
g
a co
m
p
ariso
n
p
a
ram
e
ter wh
ich
shou
ld
b
e
m
a
tch
e
d
with
th
e th
resho
l
d
ob
tain
ed
from th
e
pre
v
iously anal
yzed data.
Step 1
: in
pu
t
th
e
d
a
ta
∑
D
fr
om
t
h
e st
ora
g
e se
ve
r
Step 2
:
h
i
erarch
ically
o
r
d
e
r
th
e seg
m
en
tatio
n
ru
les with
p
r
ov
id
ed
α
,
β
,
π
,
Ω
a
n
d
dat
a
i
n
put
Step 3
:
d
e
fi
n
e
th
e test
case with in
itial sup
p
l
em
en
tin
g
p
a
ram
e
ter
α
Step 4
:
classificatio
n
based
on
th
e supp
lem
e
n
tin
g
p
a
ra
m
e
ter is d
i
v
i
ded
in to
fo
ur cl
asses C1, C
2
,
C3
, C
4
Step 5
:
C1-
refe
rs t
o
th
e classification
base
d
o
n
a
g
e
gr
o
u
p
Whe
r
e
α
–is the a
g
e
of
custom
er
β
-
i
s th
e v
a
r
i
ous ag
e gr
oup
π
-is th
e sim
ilar
ities in
sam
e
a
g
e
g
r
ou
p ratio
Ω
-d
isp
a
rities in
ag
e
g
r
ou
p in
p
u
rch
a
se
p
a
ttern
|
π
|=
∆
β
–
∆
α
,w
h
e
r
e
∆
i
s
t
h
e
devi
at
i
o
n a
n
d
|
Ω
|=
∆
β
/
∆
α
|R1
|
=|
π
|+|
Ω
|
+
|
∆
β
X
∆
α
|
|R1| corres
pondence
to
t
h
e
ratio
of t
h
e seg
m
e
n
tatio
n
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
SN
:
2
088
-87
08
I
J
ECE
Vo
l. 6
,
N
o
. 4
,
Au
gu
st 2
016
:
18
18
–
1
827
1
822
Step 6
:
C2
-refers to the classificatio
n b
a
sed
on
th
e occu
p
a
tion
Whe
r
e
α
-i
s
t
h
e occu
pa
t
i
on of
cust
om
er
β
-
i
s th
e v
a
r
i
ous o
c
cup
a
tio
n gro
up
π
-is th
e sim
ilar
ities in
th
e sam
e
o
c
cu
p
a
ti
o
n
ratio
Ω
-d
isp
a
rities in
d
e
m
o
g
r
ap
h
i
c group
in pu
rchase p
a
ttern
o
f
th
e cu
sto
m
er
|
π
|=
∆
β
–
∆
α
, whe
r
e
∆
i
s
t
h
e
devi
at
i
o
n a
n
d
|
Ω
|=
∆
β
/
∆
α
|R2
|
=|
π
|+|
Ω
|
+
|
∆
β
X
∆
α
|
|R2| corres
pondence
to
t
h
e
ratio
of t
h
e seg
m
e
n
tatio
n
Step 7
:
C3
-refers to the classificatio
n b
a
sed
on
th
e
Qu
alification
Whe
r
e
α
-i
s
t
h
e qual
i
f
i
cat
i
on of
cust
o
m
er
β
-
i
s th
e v
a
r
i
ous qu
alif
icatio
n
g
r
ou
p
π
-is th
e sim
ilar
ities in
th
e sam
e
qu
alificatio
n
ratio
Ω
-d
isp
a
rities in
b
e
h
a
v
i
oral
g
r
o
u
p
i
n
p
u
rch
a
se p
a
ttern
o
f
t
h
e cu
st
o
m
ers
|
π
|=
∆
β
–
∆
α
,w
here
∆
i
s
t
h
e
de
vi
at
i
on
an
d
|
Ω
|=
∆
β
/
∆
α
|R3
|
=|
π
|+|
Ω
|
+
|
∆
β
X
∆
α
|
|R3| corres
pondence
to
t
h
e
ratio
of t
h
e seg
m
e
n
tatio
n
Step 8
:
C4
-refers to the classificatio
n b
a
sed
on
th
e gen
d
e
r
g
r
ou
p
Whe
r
e
α
-
i
s th
e
g
e
nd
er of
cu
sto
m
er
β
-
i
s th
e v
a
r
i
ous g
e
nd
er
gr
oup
π
-is th
e sim
ilar
ities in
th
e sam
e
g
e
n
d
e
r
ratio
Ω
-d
isp
a
rities in
d
e
m
o
g
r
ap
h
i
c group
in pu
rchase p
a
ttern
o
f
th
e cu
sto
m
ers
|
π
|=
∆
β
–
∆
α
,w
he
re
∆
is t
h
e d
e
v
i
atio
n and |
Ω
|=
∆
β
/
∆
α
|R4
|
=|
π
|+|
Ω
|
+
|
∆
β
X
∆
α
|
|R4| corres
pondence
to
t
h
e
ratio
of t
h
e seg
m
e
n
tatio
n
Step 9
:
Creatin
g
a classificatio
n
ratio
facto
r
to
p
e
rfo
r
m
th
e seg
m
en
tatio
n
|R|=|R1
|
+|
R2
|+|
R
3
|
+|R4
|
St
ep 10
:
Defi
n
e
a th
resho
l
d
ratio
v
a
lu
e
wh
ich
is th
e agg
r
eg
ate of all previ
ously colle
cted data ratio
and eac
h
pr
o
duct
gr
o
u
p
has a
t
h
re
sh
ol
d
val
u
e
∆
v.
St
ep 11
: Com
p
are
the
∑
R v
a
lu
e with
∆
v
value
of eac
h product class
St
ep 12
:
Ou
t
p
u
t
th
e
ratio
v
a
l
u
e to
th
e co
rrespon
d
i
ng cla
sses (C1,
C2, C3, C4 et
c.) T
h
is has t
h
e suitable
m
a
tch obtai
ned in the
com
p
ari
s
on.
St
ep 13
:
Gen
e
rate th
e percen
tag
e
b
a
sed
an
alysis grap
h
p
l
o
ttin
g ratio v
a
lu
es an
d th
resho
l
d
v
a
lu
e.
3.
R
E
SU
LTS AN
D ANA
LY
SIS
Tabl
e 1 gi
ve
s t
h
e det
a
i
l
e
d i
n
f
o
rm
at
i
on of cu
st
om
ers wi
t
h
thei
r occ
u
pat
i
o
n, q
u
al
i
f
i
cat
i
o
n
,
age gr
o
u
p
,
g
e
nd
er and
i
n
co
m
e
o
f
t
h
e cu
sto
m
er collected
th
rough
m
u
ltip
le c
h
o
i
ce
qu
estion
s
p
r
o
v
i
d
e
d in
th
e
q
u
e
stio
nn
air
e
s. D
a
taset w
a
s co
llected
thr
ough
o
n
lin
e w
e
b
fo
r
m
s f
r
o
m
d
i
f
f
er
en
t
g
e
n
e
r
a
tion
s
to id
en
tif
y th
eir
t
a
st
es and
pre
f
erences a
b
o
u
t
m
obi
l
e
ret
a
i
l
m
a
rket
. Dat
a
s
e
t
was ri
c
h
wi
t
h
wi
de va
ri
et
y
of
dat
a
f
r
om
di
ffe
rent
incom
e
groups
as well as age groups
. To
segm
ent th
e mark
et m
a
in
ly
p
r
o
p
o
s
ed
algo
rith
m
co
n
s
iders the
constraint like age group, ge
nder, occ
u
p
a
tion
,
edu
cation
and
in
co
m
e
o
f
the family. Base
d
on
th
ese constrain
t
s
di
ffe
re
nt
p
u
rc
h
a
si
ng
pat
t
e
r
n
s
are
gene
rat
e
d
to
d
i
fferen
tiate th
e retail m
o
b
ile m
a
rk
et.
Fi
gu
re 2 p
r
ovi
des t
h
e i
n
f
o
r
m
at
i
on regar
d
i
ng
quest
i
o
n
n
ai
res aske
d i
n
t
h
e su
rvey
t
o
segm
ent
t
h
e
cust
om
er b
u
y
i
ng
pat
t
e
r
n
s
ba
sed
on
t
h
ei
r c
hoi
ces
Q
u
est
i
o
ns as
ke
d i
n
t
h
e su
rvey
were
m
a
i
n
l
y
abo
u
t
m
obi
l
e
ph
o
n
es
ho
w c
u
st
om
ers sel
ect
whi
l
e
pu
rcha
si
ng
m
obi
l
e
ph
o
n
es.
I
n
f
o
rm
at
ion
g
o
t
f
r
om
va
ri
o
u
s c
u
st
om
ers wi
t
h
di
ffe
re
nt
o
p
t
i
o
ns a
n
d
di
ffe
ren
t
pre
f
ere
n
ces.
Evaluation Warning : The document was created with Spire.PDF for Python.
IJECE
ISS
N
:
2088-8708
Seg
m
e
n
t
a
t
i
o
n
of
Ret
a
i
l
M
o
bi
l
e
Market
Usi
n
g
HMS
Al
g
o
ri
t
h
m
(
K
oyi
A
nus
ha)
1
823
Tab
l
e
1
.
C
u
stomer Details Gro
up
S.No.
Attribute
Options
Values
1 Gender
M
a
le 260
Fe
m
a
le
240
2 Age
gr
oup
15-
21
21
22-
30
167
31-
38
146
39-
45
92
Above 45
67
3
Fam
i
ly
Annual inco
m
e
5000
0-
100
000
75
1000
00-
20
000
0
200
M
o
r
e
than 200000
225
4 E
ducation
status
Co
m
puter science
149
Co
m
m
er
ce 174
LLB
2
1
MBBS
32
Other
s
124
5 Occupation
Sof
t
ware engineer
60
Bank m
a
nager
98
Law
y
er
2
3
Doctor 32
Student
82
Other
s
205
Fi
gu
re
2.
Q
u
es
t
i
onnai
r
es t
o
C
o
l
l
ect
Dat
a
set
(a)
(b
)
Fi
gu
re
3.
Ge
n
d
e
r
of t
h
e C
u
st
o
m
ers
Fi
gu
re 3
desc
ri
bes ab
o
u
t
ge
nd
er cat
ego
r
y
an
al
y
s
i
s
whi
c
h c
ont
ai
n
s
fem
a
l
e
and m
a
l
e
custom
ers i
n
t
h
e
retail
m
a
rk
et. Fig
u
re
3
(
a) is th
e statu
s
o
f
t
h
e ex
istin
g
ap
ri
o
i
ri alg
o
rith
m
b
a
sed
su
rv
ey resu
lt it sh
ows th
e resu
lt
with
th
e m
a
le
o
f
53
%
wh
ereas fem
a
le as 4
7
%
and
Figu
re 3
(
b) is th
e su
rv
ey for th
e HMS alo
g
r
ith
m
it g
i
v
e
s
th
e resu
t
o
f
m
a
le p
e
rcen
tag
e
is 52
%
wh
ere as fem
a
le
is 4
8
% in
th
e surv
ey co
ndu
ted
for m
o
b
ile p
h
o
n
e
s.Th
at
m
eans com
p
are t
o
fem
a
l
e
s
pu
rc
hasi
n
g
p
r
od
uct
s
m
a
l
e
s
are p
u
rc
hasi
n
g
m
obi
l
e
ph
on
es hi
g
h
fo
r p
e
rso
n
al
in
teractio
ns. B
o
th
t
h
e algorith
m
resu
lts are similar an
d
sh
ows
t
h
at
m
a
l
e
pu
rc
hasi
n
g
i
s
t
h
e
hi
g
h
est
c
o
m
p
ared
to
th
e
fem
a
le.so
b
o
t
h
h
a
v
e
si
milar tru
e
p
o
s
it
iv
es an
d false
p
o
s
itiv
es wh
ile tak
i
ng
g
e
nd
er
as a co
n
s
t
r
ain
.
Fi
gu
re
4
descri
bes a
g
e
gr
ou
p
of t
h
e c
u
st
om
ers b
a
sed
o
n
di
f
f
ere
n
t
scal
es.
M
obi
l
e
p
u
r
cha
s
i
ng
pat
t
e
r
n
will v
a
ry
fro
m
on
e to
ano
t
h
e
r ag
e gro
u
p
o
f
th
e cu
sto
m
er
s. Figu
re 4(a) is th
e su
rv
ey
of th
e ex
isting
ap
ri
o
r
i
alg
o
r
ith
m
,
it e
x
p
l
ains th
at the ag
e gro
u
p
betw
een
3
5
and
4
5
is th
e
gr
oup w
h
er
e th
e
pu
rch
a
sing
o
f
th
e
m
o
b
i
l
e
phones is at the highest and the ag
e gr
oup
betw
een
18
and 2
5
pu
r
c
h
a
si
n
g
is at
th
e lo
w
e
st. Fig
u
r
e
4
(
b
)
sh
ows
t
h
e res
u
l
t
of
HM
S al
g
o
ri
t
h
m
of 1
8
-
2
5
a
g
e gr
o
up s
h
ow
s
t
h
e pe
rcent
a
g
e
of
5%
w
h
i
c
h i
s
ve
ry
l
o
w
buy
i
n
g
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
SN
:
2
088
-87
08
I
JECE
Vo
l. 6
,
N
o
. 4
,
Au
gu
st 2
016
:
18
18
–
1
827
1
824
p
a
tter
n
s und
er
th
is ag
e gr
oup
w
h
er
eas
34
-45 ag
e
gr
oup
b
u
y
s m
o
r
e
m
o
b
iles as
3
6
.7%.
HMS algo
r
ith
m
sh
ow
s
th
e in
creased
resu
lt b
y
t
h
e
2% wh
en
co
m
p
ared to
ex
isting
algorith
m
.
(a)
(
b
)
Fi
gu
re 4.
A
g
e Gr
ou
p O
f
T
h
e C
u
st
om
ers
(a)
(b
)
Fi
gu
re 5.
Q
u
al
i
f
i
cat
i
on of
t
h
e
C
u
st
om
ers
Fi
gu
re 5 de
scr
i
bes t
h
e cust
o
m
er pro
f
i
l
e
st
at
us base
d o
n
t
h
ei
r j
o
bs, ex
pl
ai
ns abo
u
t
t
h
e per
f
o
r
m
a
nce
ran
g
e
of c
u
st
o
m
er pre
f
ers
w
h
at
ki
nd a
n
d
h
o
w m
a
ny
m
obi
l
e
ph
ones
.
Fi
g
u
re
5
(
a) i
s
t
h
e
resul
t
of t
h
e e
x
i
s
t
i
ng
Ap
ri
o
r
i
al
g
o
ri
t
h
m
i
t
gi
ves t
h
e re
sul
t
bet
w
een
di
ffe
rent
l
e
vel
s
s
u
ch a
s
pri
m
ary
scho
ol
,
hi
g
h
sc
ho
o
l
and
u
n
i
v
e
rsity and g
i
v
e
s t
h
e resu
lt su
ch
th
at
p
r
im
ary as th
e h
i
gh
est p
e
rcen
tag
e
and
un
i
v
ersity as th
e
lo
west.
Fig
u
re 5(b) g
i
v
e
s th
e resu
lt o
f
HMS algorith
m in
d
i
ffere
n
t
attrib
u
t
es it sh
ows th
e resu
lt o
f
p
r
im
ary sch
o
o
l
attrib
u
t
e sho
w
s th
e least
b
u
y
ing
an
d th
e
un
iv
ersity
attrib
u
t
e
pu
rch
a
ses th
e m
o
re m
o
b
ile p
hon
es.
(a)
(b
)
Fi
gu
re
6.
Occ
u
pat
i
o
n
o
f
t
h
e C
u
st
om
ers
Fi
gu
re
6
desc
ri
bes t
h
e
rat
i
o
of
t
h
e
occ
upat
i
o
n
of
t
h
e
cust
o
m
ers, t
h
e
g
r
a
p
hs e
x
pl
ai
ns t
h
e
det
a
i
l
s
o
f
t
h
e
custom
er occ
u
pation c
a
tegory accordin
g to whic
h the
what kind of occ
upie
d
custom
e
r
s
buys what kind of
m
obi
l
e
pho
nes
and
w
h
at
i
s
t
h
e rat
i
o
o
f
p
u
rc
hasi
n
g
t
h
e m
o
b
ile p
hon
es. Fi
g
u
re 6(a) is th
e resu
lt of th
e ex
isting
A
p
r
i
or
i algo
r
i
t
h
m
w
h
ich in
clu
d
e
s d
i
ff
e
r
ent
categories s
u
ch as
housewi
fe
, st
ude
nt
,
w
o
r
k
er
, e
ngi
nee
r
et
c.
according to the survey it show
s that worke
r
as the highest rate of pu
rc
has
i
ng the m
obile
phone and student is
the least. Figure 6(b) s
h
ows t
h
e re
sult of t
h
e HMS al
go
rith
m
sh
o
w
s th
e
resu
lt
o
f
t
h
e at
trib
u
t
e t
o
urism trad
e
pu
rc
hases t
h
e
m
o
st
and t
h
e
h
ous
ewi
f
e
buys
the m
obile phones t
h
e least.
Evaluation Warning : The document was created with Spire.PDF for Python.
IJECE
ISS
N
:
2088-8708
Seg
m
e
n
t
a
t
i
o
n
of
Ret
a
i
l
M
o
bi
l
e
Market
Usi
n
g
HMS
Al
g
o
ri
t
h
m
(
K
oyi
A
nus
ha)
1
825
3.
1.
Extended
fe
atures c
o
nsidere
d in HMS Al
gorithm
Inc
o
m
e
of t
h
e
cust
om
er, n
u
m
ber of m
obi
l
e
p
h
o
n
es
,
bra
n
d
of
t
h
e m
obi
l
e
, p
u
r
p
ose
of
t
h
e m
obi
l
e
phone a
n
d selection
of t
h
e
brand attributes
are c
o
nsid
ered
in th
e
p
a
p
e
r to
o
b
t
ain th
e
b
e
tter
p
r
ed
ictio
n and
pr
ofi
l
e
.
(a)
(b)
(c
)
Fi
gu
re
7.
I
n
co
m
e
s of t
h
e C
u
s
t
om
ers
Figure 7 de
scribes the income of
the customer it show
s the ratio of accordin
g to the incom
e
the
cust
om
er pu
rc
hases t
h
e m
obi
l
e
ph
ones
.
Fi
g
u
re
7(a
)
gi
ves t
h
e rat
i
o
of c
u
st
om
er who
’
s i
n
com
e
vari
es an
d h
o
w
t
h
ei
r
pu
rc
hase
t
h
e m
obi
l
e
p
h
o
n
es.
Fi
g
u
re
7(
b
)
e
xpl
ai
n
s
th
e ratio
of the fam
ily in
co
m
e
o
f
t
h
e cu
sto
m
er and
the
gra
p
h s
h
ows t
h
at the c
u
st
omer whose
fam
i
ly annual in
come
is
m
o
re
than
2
,
00
,0
00
pu
rch
a
ses
th
e m
o
b
ile
phones m
o
re and t
h
e customer whose fam
i
ly annual in
co
m
e
i
s
bet
w
een
50
,0
0
0
an
d
1,
00
,0
0
0
p
u
r
c
ha
ses t
h
e
least. Figure
7(c) e
x
plains t
h
e ratio
of the
indi
vidu
al an
nual in
co
m
e
o
f
th
e cu
st
o
m
er
an
d th
e gr
aph
s
sh
ow
s
th
at th
e cu
st
omer
wh
o
s
e indiv
i
d
u
a
l
annu
al
in
co
m
e
is b
e
t
w
een 2,
00
,0
00
an
d
3
,
60
,0
00
p
u
r
c
h
a
ses t
h
e m
o
b
ile
ph
o
n
es m
o
re a
n
d
t
h
e c
u
st
om
er w
h
ose i
ndi
v
i
dual
a
n
n
u
al
i
n
com
e
i
s
no
ne
p
u
rc
hases t
h
e
l
east
.
The
p
r
e
v
i
o
us
researc
h
pape
r descri
bes onl
y
t
h
e
i
n
di
vi
d
u
al
i
n
com
e
but
t
h
i
s
pa
per
gi
ves
t
h
e det
a
i
l
s
of
b
o
t
h
t
h
e
fam
i
ly
annual
i
n
com
e
and
i
n
di
vi
d
u
al
a
n
nua
l
i
n
com
e
whi
c
h
gi
ves
t
h
e
bet
t
er u
n
d
erst
a
ndi
ng
o
f
t
h
e c
u
st
om
er pu
rch
a
si
ng
t
h
e
m
obi
l
e
pho
nes
.
Figure
8.
Num
b
er
of the
Mobile Phones
Figure
9. B
r
a
n
d of t
h
e Mobi
le Phone
s
Fi
gu
re
8 s
h
o
w
s t
h
e i
n
fo
rm
at
ion
o
f
t
h
e c
u
st
om
ers havi
ng
t
h
e m
obi
l
e
ph
one
s.
X-a
x
i
s
i
n
t
h
e
gra
p
h
expl
ai
n
s
h
o
w
m
a
ny
pho
nes
doe
s t
h
e i
ndi
vi
dual
pe
rs
on c
o
nsi
s
t
of
, w
h
ere
a
s Y-a
x
i
s
ex
pl
ai
ns t
h
e pe
rent
age o
f
t
h
e m
obi
l
e
p
h
o
n
es.
M
o
st
of
t
h
e c
u
st
om
ers a
r
e c
h
o
o
se
d
onl
y
one
m
obi
l
e
ph
o
n
e
fo
r
usa
g
e wi
t
h
pe
rcent
a
ge
o
f
52
.2
0%
. Less
num
ber
of c
u
s
t
om
ers are sel
ect
ed m
o
re t
h
a
n
3
p
h
ones
wi
t
h
pe
rce
n
t
a
ge
o
f
5
.
6
0
%.
R
e
m
a
i
n
i
n
g
cust
om
ers cho
o
se
d 2 an
d
3 m
obi
l
e
pho
nes
wi
t
h
perce
n
t
a
ge o
f
3
1
% an
d
11.
2
0
% t
o
use
t
h
e
m
obi
l
e
ph
one
fo
r
their pu
rp
oses
.
Figu
re
9
tells
ab
ou
t
d
i
fferen
t
categ
o
r
ies
m
o
b
ile p
hon
es.
X-ax
is in
th
e
g
r
ap
h
tells abo
u
t
th
e d
i
fferen
t
categ
or
ies o
f
t
h
e m
o
b
ile p
h
o
n
e
s su
ch
as No
k
i
a, Sam
s
u
ng, I
P
hon
e, Sony an
d
o
t
h
e
r
s
,
w
h
er
eas Y-
ax
is tells
abo
u
t
t
h
e perc
ent
a
ge
o
f
w
h
i
c
h
cat
e
g
o
r
y
of
m
obi
l
e
ph
one
i
s
bee
n
p
u
rc
ha
sed.
C
o
nsi
d
e
r
e
d
t
h
e
s
u
r
v
ey
t
h
ere
i
s
vary
i
n
g pe
rce
n
t
a
ge of
di
f
f
ere
n
t
cat
ego
r
i
e
s o
f
m
obi
l
e
phone
s being purcha
s
ed in the cas
e custom
ers are
m
o
stly
pre
f
er
red sam
s
un
g p
h
ones
wi
t
h
per
cent
a
ge o
f
28
.6
0% c
o
m
p
are t
o
ot
her c
a
t
e
go
ri
es of m
obi
l
e
p
h
o
n
es a
nd s
ony
is th
e least
p
u
rch
a
sing
con
s
ider
ed w
ith th
e
oth
e
r
categ
or
ies
o
f
m
o
b
ile p
hones.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
SN
:
2
088
-87
08
I
JECE
Vo
l. 6
,
N
o
. 4
,
Au
gu
st 2
016
:
18
18
–
1
827
1
826
Figure
10.
Purpose
of
usi
n
g Mobile P
h
ones
Figure
11. Selection
of the
Bra
n
d
Fi
gu
re
10
ex
pl
ai
ns f
o
r
w
h
at
p
u
r
p
ose t
h
e m
o
b
ile ph
on
es is
b
een u
s
ed
, th
ere are se
veral
purposes
for
whi
c
h m
obi
l
e
ph
o
n
es i
s
bee
n
use
d
acc
or
di
ng t
o
t
h
e i
n
di
vi
d
u
al
cust
om
er be
havi
ou
r,
cust
om
er
m
o
st
l
y
uses
m
obi
l
e
pho
nes
t
o
di
ffe
rent
p
u
r
p
o
se l
i
k
e m
u
l
tim
edi
a
,cam
era,cal
l
s
/
s
m
s
and ot
he
r uses. C
o
nsi
d
e
r
ed t
h
e su
rvey
th
e pur
po
se
of u
s
i
n
g a m
o
b
ile ph
on
e is mu
lti
m
e
d
i
a
with con
s
istin
g of
3
1
.20
%
co
m
p
are to o
t
h
e
r op
tio
ns
gi
ve
n i
n
t
h
e
s
u
r
v
ey
t
o
c
u
st
o
m
ers.
Second
selected option by c
u
stom
ers
was calls/sms with
p
e
rcen
tag
e
of
30
.4
0%
, rem
a
ini
n
g
opt
i
o
ns
w
e
re c
onsi
s
t
i
n
g
of
l
e
ss
perce
n
t
a
ge c
o
m
p
are t
o
t
h
ese
o
p
t
i
ons
.
Fig
u
re
1
1
i
n
trep
rets ab
ou
t th
e rang
e for selectio
n
of
m
o
b
ile p
hon
es
wh
ich
in
cludes l
o
w, avera
g
e a
n
d
hi
g
h
. M
o
st
o
f
t
h
e cust
om
ers prefe
rre
d m
obi
l
e
ran
g
e hi
g
h
c
o
st
wi
t
h
4
7
%.
R
e
m
a
i
n
i
ng 2
7
%
cust
om
ers p
r
efe
rre
d
avera
g
e c
o
st
m
obi
l
e
s a
nd
bel
o
w 2
5
% c
u
st
om
ers ch
o
o
se
d l
o
w ra
n
g
e m
obi
l
e
s t
o
use t
h
ei
r
pers
o
n
al
i
n
t
e
ra
ct
i
ons.
There
is
no m
u
ch
diffe
re
nce
between l
o
w ra
nge a
n
d
av
erag
e ran
g
e
m
o
b
iles to
purch
ase m
o
b
ile ph
on
es.
Fi
gu
re
1
2
. C
o
m
p
ari
s
ons
bet
w
een
A
p
ri
ori
a
n
d
HM
S
Al
go
r
i
t
h
m
Figure 12
expl
ains
the ratio between Apriori
al
go
ri
t
h
m
and
HM
S al
go
ri
t
h
m
.
The abo
v
e
gra
p
h gi
ves
t
h
e det
a
i
l
e
d rat
i
o bet
w
ee
n
di
f
f
ere
n
t
cat
eg
ori
e
s suc
h
as A
g
e, Ge
nde
r,
Q
u
al
i
f
i
cat
i
on, Oc
cupat
i
o
n a
nd
I
n
com
e
,
th
e g
r
aph
exp
l
ain
s
th
at th
e HMS alg
o
rith
m
g
i
v
e
s th
e
b
e
tter resu
lt co
m
p
ared
to
t
h
e Ap
ri
o
r
i algo
rith
m
i
n
th
e
case of ab
ov
e categ
o
ries tak
e
n in
to
con
s
id
eratio
n
.
4.
CO
NCL
USI
O
N
Anal
y
z
i
n
g an
d
pre
d
i
c
t
i
ng t
h
e pu
rchase
pat
t
erns an
d va
ri
ous c
o
nst
r
ai
nt
s i
n
t
e
gral
t
o
r
e
t
a
i
l
m
obi
le
mark
et u
s
i
n
g
h
ybrid
m
a
rk
et seg
m
en
tatio
n
(HMS) algo
rit
h
m
is b
e
in
g
carried ou
t th
e
d
a
taset was collected
fr
om
vari
ous a
g
e g
r
ou
ps.
The
resul
t
s
obt
ai
n
e
d s
u
p
p
o
rt
ed t
h
e p
r
op
ose
d
m
e
t
h
o
dol
ogy
wi
t
h
an i
m
pro
v
e
d
resul
t
com
p
ared t
o
t
h
e exi
s
t
i
ng al
g
o
r
i
t
h
m
s
used i
n
segm
ent
a
t
i
on. Im
provi
ng t
h
e
dat
a
set
po
p
u
l
a
t
i
on an
d ad
di
n
g
m
o
re
con
s
t
r
ai
nt
s can
im
prove t
h
e
per
f
o
r
m
a
nce of t
h
e al
gori
t
h
m
. The al
gori
t
hm
can be general
i
zed a
nd
use
d
i
n
vari
ous
ret
a
i
l
a
n
d
w
h
ol
esal
e
m
a
rket
t
o
get
b
e
t
t
e
r segm
ent
a
t
i
on a
n
d
pu
rc
ha
se pat
t
e
r
n
s.
REFERE
NC
ES
[1]
S. S. Gulluoglu,
“Segmenting custom
ers with data mining Techn
i
ques,”
IEEE. Th
ird International Conference
, pp.
154-159, 2015
.
[2]
B. M. Ramageri
and B. L.
Des
a
i,
“Role of data
mining in retail Sector,”
In
ternation
a
l Journal on Computer Scien
c
e
and Engin
eering
(
I
JCSE)
, vol/iss
u
e: 5(01)
, 2013
.
[3]
A.
A.
Raorane,
et al.
, “Association rule– extracting knowledge
using market basket analy
s
is
,”
Research Journal of
Recent S
c
iences
, vol/issue: 1(2)
,
pp. 19-27
, 2012
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
J
ECE
I
S
SN
:
208
8-8
7
0
8
Seg
m
e
n
t
a
t
i
o
n
of
Ret
a
i
l
M
o
bi
l
e
Market
Usi
n
g
HMS
Al
g
o
ri
t
h
m
(
K
oyi
A
nus
ha)
1
827
[4]
M.
J.
Sha
w
,
et al.
, “Knowledge
management an
d
data mining for marketing,”
Decision support systems
, vol/issue:
31(1), pp
.127-13
7, 2001
.
[5]
A. P
a
rvaneh
,
et al.
,
“Combining data
mining and group
decision
ma
king in retailer segmentation
based on LRFMP
variab
les
,
”
J Ind
Eng
Prod Res,
v
o
l/issue: 2
5
(3), p
p
. 197-206
, 201
4.
[6]
M.
Y.
Avcilar and E. Yakut,
“
A
ssoci
ation Ru
les in Data Mining: An A
pplication on
a Clo
t
hing and A
ccesso
r
y
S
p
ecia
l
t
y
S
t
or
e,
”
Canadian So
cia
l
Science
, vol/issue: 10(3)
, pp
.75-
83, 2014
.
[7]
S.
Sure
sh,
et
al.
,
“Application of
Retail Analy
tic
s Using Association Rule Min
i
ng
In Data Min
i
ng
Techn
i
ques With
Res
p
ect
To
Ret
a
il S
uperm
arke
t,
”
IJEM
R,
vol. 5
,
2015.
[8]
P. Q. Brito,
et al.
, “
C
ustom
e
r segm
entation in a l
a
rge datab
a
se
of an online customized fashion business,”
Robotics
and Computer-I
ntegrated
Manufacturing,
2015.
[9]
A.
Joshi,
et al.
, “Use of Data Mining Techn
i
ques to Improve
the Eff
ect
iv
enes
s
of S
a
les
and M
a
rketing
,
”
International Jo
urnal of Co
mputer Science and
Mobile
Computing (
I
JCSMC)
, vol. 4
,
pp
. 81-87
,
2015.
[10]
V. L. Migueis,
et al.
, “Customer data mining for lifesty
le Segmentation
,
”
Expe
rt Sy
ste
m
s wi
th Applications
,
vol/issue:
39(10)
, pp
. 9359-9366
, 2012.
[11]
B. Noori
and
W
.
T
.
Br
an
ch, “
A
n Anal
ys
is
of
Mobile Bank
ing User Beh
a
vior
Using Customer
Segm
entation
,
”
I
n
ternational Jou
r
nal of Globa
l
Business
, vol/issue: 8(2)
, 2015
.
[12]
B. K. Bar
a
dwaj
and S. Pal,
“Mining education
a
l data
to an
aly
ze students'
p
e
rfor
m
ance,”
Interna
tional Journal o
f
Advanced Comp
uter Science
and
Applications (
I
JACSA)
, vol/issue: 2(6), 2011.
[13]
X. Liu
and H.
Liu, “An impro
v
ed apri
ori
algo
rithm
for associ
ation ru
le
,”
TEL
K
OMNIKA Indonesian Journal
of
Ele
c
trica
l
Eng
i
n
eering
, vol.11,
pp. 6521-6526
,
2013.
[14]
Q. Zhou,
et
al
., “Intelligent data mini
ng
and decision s
y
s
t
em fo
r co
mmercial decision making
,”
TE
LKOMNIKA
Indonesian Jour
nal of El
ectrical Engineering,
vol. 12
, pp
. 792-80
1, 2014
.
[15]
J. Hipp,
et al.
, “Algorithms f
o
r association r
u
le mi
ning—a general
survey
and
comparison,”
ACM sigkdd
explorations
ne
wsletter
, vol/issue: 2(1)
, pp
. 58-6
4
, 2000
.
Evaluation Warning : The document was created with Spire.PDF for Python.