Indonesi
an
Journa
l
of El
ect
ri
cal Engineer
ing
an
d
Comp
ut
er
Scie
nce
Vo
l.
12
,
No.
3
,
Decem
ber
201
8
, p
p.
103
7
~
10
44
IS
S
N: 25
02
-
4752, DO
I: 10
.11
591/ijeecs
.v1
2
.i
3
.pp
103
7
-
10
44
1037
Journ
al h
om
e
page
:
http:
//
ia
es
core.c
om/j
ourn
als/i
ndex.
ph
p/ij
eecs
Classific
atio
n
o
f
t
he Main
stay Ec
on
omic R
eg
i
on Us
ing Deci
sion
Tree Me
thod
Heru Is
ma
nto
1
,
Az
ha
ri
Az
ha
ri
2
,
S
uh
arto
Suh
arto
3
,
Li
n
colin Ar
syad
4
1
,2,3
Depa
rtment
o
f
Com
pute
r
Sci
e
nce
and El
e
ct
ron
ic
s,
Fa
cul
t
y
of
Mathe
m
at
i
cs
an
d
Natur
a
l
Sc
ie
nc
e,
Univer
sita
s Gad
j
ah
Mada
,
Yog
y
a
kar
ta,
Indon
esia
1
Depa
rtment of I
nform
at
ic
s E
ng
i
nee
ring
,
Fa
cul
t
y
of
Engi
n
ee
rin
g,
Mus
amus
Univer
sit
y
,
Mer
auke,
I
ndonesia
4
Depa
rtment of
Ec
onom
ic
s,
Fac
ulty
of
E
conomics
and
Busin
ess,
Univer
sit
as
Gad
ja
h
Mad
a, Yog
y
aka
rt
a, I
ndonesi
a
Art
ic
le
In
f
o
ABSTR
A
CT
Art
ic
le
history:
Re
cei
ved
J
un
1
9
, 2
018
Re
vised
A
ug
10, 2
018
Accepte
d
Aug
25, 201
8
The
dev
el
opm
e
nt
of
the
reg
io
n
ca
nnot
be
se
par
ated
from
th
e
concept
o
f
ec
onom
ic
growt
h
and
the
de
te
r
m
ina
ti
on
of
the
m
ai
nsta
y
r
egi
on
as
a
reg
ion
a
l
ce
nt
er
th
at
is
ex
pec
t
ed
to
hav
e
a
positi
ve
impac
t
on
e
conomic
gr
owth
to
th
e
surrounding
reg
i
ons.
In
fa
ct,
th
e
d
eterm
ina
ti
on
of
the
m
ai
nsta
y
reg
ion
is
a
diffi
cu
lt
thi
ng
to
do.
Som
e
c
ase
s
of
th
e
de
te
rm
in
at
ion
o
f
th
e
m
ains
tay
reg
ion
are
m
ostl
y
on
th
e
basis
of
th
e
pr
ero
gative
r
ight
s
of
the
pol
icy
m
a
ker
s
without
ca
ref
u
lly
se
ei
ng
the
a
chieve
m
e
nts
of
the
dev
e
lopment
of
a
r
egi
on.
The
obje
c
ti
ve
of
thi
s
stud
y
is
to
dev
el
op
a
cl
assifi
cation
m
odel
of
th
e
m
ai
ns
t
a
y
ec
onom
ic
reg
ion
using
computat
i
onal
techniqu
es.
The
decision
tr
ee
m
et
hods
of
NBTre
e
and
J48
are
used
in
thi
s
stud
y
a
nd
combined
with
Klassen
t
y
po
log
y
.
Th
e
r
esult
s
of
t
his
stud
y
show
tha
t
J48
al
gorit
hm
has
bet
te
r
ac
cur
acy
tha
n
NBTre
e
in
the
form
at
ion
pro
ce
ss
of
de
ci
sio
n
tre
e
.
Th
e
ac
cur
acy
of
J48 i
s hi
gher
tha
n
NB
Tre
e i.
e
.
68.
96%
.
The
compara
t
iv
e
result
of
the
class
ifi
cation
of
the
m
ai
nsta
y
ec
onom
ic
reg
io
n
bet
wee
n
Klassen
and
J48
show
s
tha
t
the
re
is
a
shift
in
the
cl
ass
positi
on
of
the
dev
el
opm
en
t
quadr
an
t
.
In
Klassen
cl
assi
fic
a
ti
on,
the
r
e
ar
e
three
reg
ions
t
hat
ar
e
c
at
egor
ized
int
o
the
m
ai
nsta
y
r
egi
o
ns
with
adva
nce
d
developm
e
nt
and
rap
id
g
rowth
(K1).
Mea
nwhile,
J48
result
s
show
tha
t
the
r
e
is
no
reg
ion
cate
gori
z
ed
int
o
K1.
How
eve
r,
th
e
m
ai
nsta
y
e
cono
m
ic
reg
ion
on
J48
is
base
d
on
the
le
v
el
o
f
deve
lopment
wit
h
the
le
v
el
b
el
o
w
K1,
i.e.
K2.
J
48
cl
assifi
cation
result
s
show
tha
t
the
r
e
are
t
e
n
reg
enc
i
es
tha
t
are
cate
gori
ze
d
i
nto
t
he
m
ai
nsta
y
ec
onom
ic
reg
ions,
namel
y
Bia
k,
Reg
ency
of
Ja
y
apur
a,
Ja
yawij
a
y
a
,
Kerom
,
Mera
uke
,
Mim
ika
,
Nabi
re,
Ndunga, Yape
n
,
and the Munici
p
al
ity
of
J
a
y
apur
a
.
Ke
yw
or
ds:
Decisi
on T
ree
J4
8
Klassen
Ma
instay
r
egi
on
NBTree
Copyright
©
201
8
Instit
ut
e
o
f Ad
vanc
ed
Engi
n
ee
r
ing
and
S
cienc
e
.
Al
l
rights re
serv
ed.
Corres
pond
in
g
Aut
h
or
:
Heru
Ism
anto,
Dep
a
rtm
ent o
f C
om
pu
te
r
Scie
nce a
nd Elec
tr
on
ic
s
,
Faculty
of Mat
hem
atics and
Natu
ral Scie
nc
e,
Un
i
ver
sit
as
Ga
dj
a
h
Ma
da, Y
ogya
ka
rta,
I
ndonesi
a.
Em
a
il
:
her
u@u
nm
us
.ac.id
1.
INTROD
U
CTION
The
m
ai
ns
ta
y
econom
ic
reg
ion
is
a
n
reg
i
on
us
e
d
a
s
a
ba
rom
et
er
of
the
e
c
onom
ic
gr
owth
of
a
re
gi
on
so
that
it
becom
es
the
econo
m
ic
su
ppor
t
f
or
oth
e
r
re
gions.
T
he
dete
r
m
inati
on
of
th
e
m
ai
ns
ta
y
ec
onom
ic
reg
i
on
is u
s
ually
co
nducted by l
oo
ki
ng
at the
achievem
ent
s o
f
the r
el
e
va
nt r
egi
on
al
d
e
vel
op
m
ent b
ased
on
the
data
of
gross
r
egio
nal
dom
es
ti
c
sect
or
(G
R
DP
)
.
T
her
e
a
re
so
m
e
app
r
oac
hes
use
d
to
det
erm
ine
the
m
a
instay
econom
ic
reg
ion
;
one
of
th
e
m
is
Klassen
ty
po
lo
gy.
Kl
assen
ty
polo
gy
cl
assifi
es
th
e
re
gions
int
o
f
our
dev
el
op
m
ent
quad
ra
nts.
Quad
ran
t
I
is
a
dev
e
lop
e
d
a
nd
ra
pi
d
gro
wth
re
gions;
Qu
a
dr
a
nt
I
I
is
a
n
a
dvance
d
but
depresse
d
reg
i
on,
Q
ua
dr
a
nt
I
II
is
a
pote
ntia
l
or
de
velo
ping
re
gion;
a
nd
Qu
a
drant
IV
i
s
a
relat
ively
la
gg
i
ng
reg
i
on
[1
]
.
By
seei
ng
a
n
re
gi
on
cat
e
gorize
d
into
a
pa
rtic
ular
de
velo
pm
ent
qu
a
dr
a
nt,
the
reg
io
n
wh
ic
h
is
the
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2502
-
4752
Ind
on
esi
a
n
J
E
le
c Eng &
Co
m
p
Sci,
Vo
l.
12
, N
o.
3
,
Dece
m
ber
2
01
8
:
103
7
–
10
44
1038
m
ai
ns
ta
y
econom
ic
reg
ion
ca
n
be
i
den
ti
fie
d.
The
re
gions
c
at
egorized
i
nto
Q
uadra
nt
I
ar
e
usual
ly
us
e
d
as
the
m
ai
n
m
a
instay
eco
no
m
ic
regi
on
by
t
he
loc
al
gove
rn
m
ent,
the
quad
ran
t
-
I
I
re
gion
bec
o
m
es
the
sec
ond
-
le
ve
l
m
ai
ns
ta
y
econom
ic
reg
ion
,
w
her
eas
t
he
re
gio
ns
cat
e
gorize
d
into
Qu
a
dran
t
II
I
a
nd
IV
a
re
no
t
cat
eg
or
iz
e
d
as
a
m
ai
ns
ta
y ec
onom
ic
r
egion.
It m
eans th
at
those re
gions s
ho
uld
be p
rior
it
iz
ed
in
fur
t
her de
velo
pm
ent act
i
viti
es.
The
or
et
ic
al
ly
,
Klassen
is
abl
e
to
identify
the
m
ai
ns
ta
y
econ
om
ic
reg
ion
base
d
on
the
r
esults
of
th
e
Gross
Re
giona
l
Do
m
est
ic
Pr
oduct
(
GRD
P)
sect
or
data
cl
us
te
rin
g
by
lo
ok
i
ng
at
the
de
velo
pm
ent
quadr
a
nt
form
ed.
Howe
ver,
the
sta
ges
of
cl
us
te
rin
g
ar
e
ver
y
rigid
a
nd
do
not
pa
y
attention
to
the
c
har
act
erist
ic
s
of
the
data
an
d
the
di
sta
nce
betwee
n
the
data
of
it
s
GRDP
[
2].
In
a
dd
it
io
n,
th
e
cl
us
te
rin
g
of
the
m
ai
ns
ta
y
reg
i
on
with
Klasse
n
al
ways
sel
ect
s
the
overall
at
tribu
te
s
of
GR
DP
sect
or
dat
a
owne
d
by
a
reg
i
on
a
s
a
whole.
Ther
e
f
or
e,
it
ta
kes
m
uch
tim
e
to
cl
assify
t
he
m
ai
ns
ta
y
e
conom
ic
reg
ion
.
T
his
stud
y
was
co
nducte
d
in
an
at
tem
pt
to
pro
vid
e
an
al
te
rnat
ive
ap
proac
h
to
cl
assi
fy
th
e
m
ai
ns
ta
y
ec
onom
ic
reg
io
n
us
i
ng
de
ci
sio
n
tre
e
com
pu
ta
ti
on
t
echn
i
qu
e
.
Dec
isi
on
T
ree
is
form
ed
from
a
s
et
of
dat
a
that
f
or
m
a
sm
aller
subs
et
and
interco
nnect
be
tween
one
to
ano
t
her
at
tri
bute
wh
ic
h
f
or
m
decisi
on
tree
st
ru
ct
ur
e.
I
n
the
process
of
f
orm
ing
a
decisi
on
tree,
it
need
the
cal
culat
ion
of
gain
val
ue
to
di
vi
de
data
with
the
sam
e
or
si
m
il
ar
instances
i
nto
sm
a
ll
er
su
bs
et
s.
A
fter
wards,
the
gain
val
ue
cal
culat
io
n
res
ult
is
us
e
d
to
cal
culat
e
the
e
ntr
op
y
value
.
This
entr
opy
value
is
us
e
d
to
determ
ine
wh
ic
h
pr
im
ary
at
tribu
te
s
are
sel
ect
ed
as
the
determ
inant
of
data
cl
assifi
cat
ion
,
fo
ll
owe
d
by
ot
her
at
tribu
te
s
that
are
ar
range
d
acc
ordin
g
t
o
their
e
ntr
op
y
values
.
I
n
cl
ust
erin
g
the
m
a
instay
econom
ic
reg
io
n
us
i
ng
Klasse
n,
al
l
at
tribu
te
s
are
seen
a
s
the
sam
e.
Me
anw
hile,
w
hen
it
use
s
the
decisi
on
tree
,
there
is
a
sel
ect
ion
of
cl
as
sifie
d
determ
i
nan
t
at
tri
bu
te
s
that
are
s
or
t
ed
by
e
ntropy
value
cal
culat
ion
s
re
peatedly
, so it
can
no
t
be d
one u
si
ng K
la
sse
n.
2.
DECISIO
N
T
REE
Decisi
on
tree
is
on
e
of
the
da
ta
cl
assifi
cat
i
on
te
c
hn
i
qu
e
s
that
m
akes
de
ci
sion
tree
str
uctu
re
m
or
e
easi
ly
un
de
rstood
[
3].
Each
i
nt
ern
al
no
de
re
presents
te
sti
ng
of
a
n
at
trib
ute,
each
bra
nch
r
epr
ese
nts
ou
t
put
of
the
te
sti
ng
,
an
d
the
le
af
no
de
represe
nts
cl
asses
or
cl
ass
distribu
ti
ons
[
4]
-
[5
]
.
T
he
to
pm
os
t
node
is
cal
le
d
r
oot
node.
T
he
root
node
will
have
so
m
e
exiti
ng
ed
ges,
but
it
does
not
ha
ve
a
n
in
com
ing
e
dge.
The
inter
na
l
node
will
hav
e
one
i
nco
m
ing
e
dg
e
and
so
m
e
exiti
ng
ed
ges.
Me
anwhil
e,
the
le
a
f
no
de
will
on
l
y
hav
e
one
inc
om
ing
edg
e
and
no e
xi
ti
ng
edge.
The
decisi
on
t
ree
is
us
e
d
to
cl
assify
an
un
known
cl
ass
s
a
m
ple
into
e
xi
sti
ng
cl
asses
.
The
da
ta
te
st
path
will
firstly
go
th
rou
gh
the
root
no
de
and
finall
y
go
throu
gh
t
he
le
af
node
t
hat
will
infer
the
cl
ass
pr
e
dicti
on
s
of
t
he
data.
T
he
da
ta
at
tribu
te
m
us
t
be
a
cat
egorical
data;
if
it
is
con
ti
nuous,
the
at
tribu
te
m
us
t
be
discreti
zed
f
irs
t [4
]
.
Thi
s techn
i
qu
e
is w
idely
u
sed
fo
r
cl
assifi
cat
ion
of
stu
den
t e
xam
p
assing
grade [
6, 7] iden
t
ific
at
ion
of
the
risk
of
tra
um
a
in
childbirt
h
thr
ough
patie
nt
data
cl
assif
ic
at
ion
[
8]
as
well
as
the
cl
assifi
cat
ion
of
r
egio
nal
dev
el
op
m
ent level [
9]. T
he fol
lowings
ar
e t
he
explanat
io
n of NBT
ree a
nd J
48 tech
niques
us
e
d
in
this st
udy.
2
.
1.
NBTre
e
Algori
th
m
NBTree
use
s
the
fr
e
quency
of
a
cl
ass
appearin
g
in
the
f
or
m
at
ion
of
a
decisi
on
tree
f
ro
m
a
set
of
data.
A
stu
dy
[
10
]
sta
te
s
that
the
NBT
ree
al
gorithm
us
es
the
Naive
Ba
ye
s
m
e
tho
d
to
de
te
rm
ine
a
le
av
e
tree
wh
il
e
gen
e
rati
ng a
decisi
on tr
ee. Bel
ow are
t
he NBTree
alg
or
it
hm
s:
a)
Determ
ine the
init
ia
l condit
ion
s.
b)
Cl
assify
the d
a
ta
an
d ca
lc
ulat
e the
value
of s
pited
node.
c)
Trim
the tree that h
a
s
been f
o
rm
ed
to ev
al
ua
te
the optim
al
t
ree a
nd cr
os
s
-
va
li
dation
e
rror.
d)
Try it
out usi
ng the
test
d
at
a
of the tree
a
nd
identify
the
ter
m
inal node
bas
ed on t
he
te
st
da
ta
.
e)
Pr
e
dict o
ne
ste
p
a
head u
sin
g Naive Bay
es at
the
gen
e
rated
term
inal no
de
.
By
assigning
a
set
of
instance
s
to
a
node,
th
e
NBTree
al
go
rithm
wil
l
evaluate
the
util
it
y
of
sp
li
t
for
each
at
trib
ute. I
f
t
he
la
r
gest u
t
il
ity
of
al
l
at
tri
bu
te
s
is
hi
gh
e
r
than
t
he
util
it
y
of
the
c
urre
nt n
ode,
the d
ivis
ion
of
existi
ng insta
nc
es w
il
l be
b
as
ed on t
hose att
r
ibu
te
s
[11].
The
util
it
y
of
node
is
cal
culat
ed
by
discreti
z
ing
the
existi
ng
data
an
d
cal
culat
ing
the
est
im
at
ion
of
5
-
fo
l
d
cro
s
s
validat
ion
acc
ur
ac
y
of
the
naïve
-
bayes
us
a
ge
a
t
the
no
de
.
Me
anwhil
e,
the
util
it
y
of
sp
li
t
is
the
weig
hted
am
ou
nt
of
the
util
ity
of
no
de
,
w
he
re
the
wei
gh
ts
assigne
d
to
a
node
a
re
pro
por
ti
on
al
to
the
num
ber
of
insta
nces d
e
rive
d
by
that
node
.
The
div
isi
on
is
set
up
sig
nificantl
y
if
th
e
relat
ive
redu
ct
ion
to
er
ror
is
bette
r
than 5%
and t
he
re ar
e
at le
ast
30 insta
nces in t
he node.
T
his is to
a
void a
ny
d
ivisi
on b
y sm
al
l values
[11].
An
NBT
ree
cl
assifi
er
s
pecifi
es
the
cl
ass
la
bel
of
an
insta
nce
by
s
ort
ing
it
into
a
le
af
and
a
pply
ing
Naïve
-
Ba
ye
s
in
the
le
af.
T
he
NBT
ree
of
te
n
achie
ves
a
hi
gh
e
r
de
gr
ee
of
accu
racy
w
he
n
c
om
par
ed
t
o
Naïve
Ba
ye
sia
n
cl
as
sifie
r [12]
.
Evaluation Warning : The document was created with Spire.PDF for Python.
Ind
on
esi
a
n
J
E
le
c Eng &
Co
m
p
Sci
IS
S
N:
25
02
-
4752
Cl
as
sif
ic
ation
of
the
Mai
ns
ta
y Eco
nomic R
e
gion
Using
De
ci
sion
Tree Me
thod
(
Her
u Is
m
an
t
o
)
1039
2
.
2
.
J48 Alg
or
ithm
The
J
48
al
gorithm
is
the
res
ul
t
of
t
he
de
velop
m
ent
of
ID3
te
ch
nique
[13
]
and
the
deter
m
inati
on
of
the
decisi
on
tr
ee
root
is
cond
ucted
by
looki
ng
at
the
gain
and
t
he
rati
o
of
the
gai
n
of
a
n
at
trib
ute.
Be
low
a
re
the J
48 alg
or
it
hm
s:
a)
Sele
ct
an
att
rib
ute as a
r
oo
t
b)
Creat
e a bra
nc
h for eac
h valu
e
c)
Divid
e
the c
as
es to t
he bra
nc
hes
d)
Re
peat the
pro
cess f
or each
branc
h
s
o
t
hat al
l t
he
cases
on t
he bra
nc
h hav
e
the sam
e cla
ss
The
J
48
al
gori
thm
ign
ores
t
he
m
issi
ng
val
ue
,
i.e.
a
value
for
a
pre
dicta
bl
e
it
e
m
based
on
w
h
at
is
known
a
bout
the
at
tribu
te
val
ues
in
the
ot
he
r
row.
T
he
basi
c
idea
of
this
al
gorithm
is
to
div
ide d
at
a
into
ra
nge
base
d
on
the
a
tt
ribu
te
val
ues
for
it
e
m
s
fo
und
i
n
trai
ning
data
set
s.
The
J4
8
al
gorithm
al
lows
cl
assifi
cat
ion
ei
ther
th
rou
gh
decisi
on tree
s
or rule
s
gen
e
ra
te
d
f
ro
m
the fo
rm
ation
of
clas
sifie
r
[
14]
.
3.
MA
I
NS
T
A
Y REGIO
N
Ma
instay
reg
i
on
is
a
n
re
gi
on
with
gre
at
er
eco
no
m
ic
gro
wth
po
te
ntial
com
par
ed
to
oth
e
r
reg
i
on
s
[
15
]
.
This
eco
nom
i
c
gro
wth
is
usual
ly
determ
i
ned
by
th
ree
i
m
po
rtant
factor
s
,
nam
el
y:
capit
al
accum
ulati
on
,
popula
ti
on
gro
wth,
a
nd
te
chnolo
gical
adv
a
nc
e
m
ent
owned
by
a
reg
io
n
[
16]
.
The
e
xisten
ce
of
m
ai
ns
ta
y
reg
io
n
is
ex
pected
t
o
ha
ve
a
posit
ive
im
pa
ct
on
t
he
eco
no
m
ic
gr
owth
f
or
oth
er
reg
io
ns
surr
ou
nd
i
ng.
So
far
,
t
he
determ
inati
on
of
m
ai
ns
ta
y
reg
io
n
is
usual
ly
co
nducted
by
the
gove
rn
m
ent
thr
ough
the
dec
isi
ons
set
fo
rth
in
th
e
Nati
on
al
Spat
ia
l
Plann
ing
Law
[15].
H
ow
e
ve
r,
it
can
al
so
be
deter
m
ined
based
on
the
cl
assifi
cat
ion
of
dev
el
op
m
ent r
egi
on
s
usi
ng
Klassen
app
ro
a
ch [1].
4.
KLASSE
N
T
YPOLO
GY
Klassen
ty
polo
gy
is
an
app
r
oa
ch
us
e
d
to
look
at
the
patte
rn
of
the
eco
no
m
ic
dev
el
opm
ent
gr
owth
of
a re
gion [1
7].
Klassen
d
i
vid
e
s the
reg
i
ons in
to fo
ur
de
velo
pm
ent
qu
a
drants
as s
how
n
in
T
able 1.
Table
1.
Cl
assi
ficat
ion
of Ec
onom
ic
G
r
ow
t
h by Klasse
n Ty
po
l
og
y
Qu
ad
rant I
(K1)
d
ev
elo
p
ed
an
d
f
a
st
-
g
rowin
g
regio
n
s
Qu
ad
rant II
(K2)
d
ev
elo
p
ed
b
u
t
d
ep
ressed
regio
n
s
Qu
ad
rant I
II
(K3)
Po
ten
tial
o
r
d
ev
elo
p
in
g
regio
n
s
Qu
ad
rant IV
(K4)
relatively
lagg
in
g
r
eg
io
n
s
Adva
nced
a
nd
rap
idly
gro
w
ing
sect
or
(
de
velo
ped
sect
or)
is
in
Qu
a
dr
ant
I.
T
his
qu
adr
a
nt
is
a
qu
a
drant
of
a
sp
eci
fic
sect
or
gr
owth
rate
in
GRD
P
(si)
wh
ic
h
is
gr
eat
er
than
the
sec
tor
gro
wth
rat
e
in
the
reg
i
on
al
GRD
P
as
the
re
fer
e
nce
(s
)
a
nd
ha
s
a
sect
or
c
ontrib
ution
value
to
GRD
P
(s
ki)
wh
ic
h
is
great
er
th
a
n
the
sect
or
c
ontrib
ution
t
o
re
gio
nal
GRDP
a
s
the
ref
e
rence
(sk).
T
his
cl
as
sific
at
ion
is
de
no
te
d
with
si>
s
an
d
sk
is>
s
k.
Adva
nced
but
sta
gn
a
nt
sect
or
is
in
Q
ua
drant
II.
This
quad
ra
nt
is
a
qu
ad
ra
nt
of
a
s
pecific
sect
or
grow
t
h
rate
in
GRDP
that
is
s
m
al
le
r
than
t
he
sect
or
gro
wth
rate
in
the
re
gio
nal
GR
DP
as
the
re
fer
e
nce
(
s)
but
has
great
er
se
ct
or
co
ntri
bu
ti
on
value
to
G
RDP
(s
ki)
tha
n
the
sect
or
con
t
rib
ution
to
r
egi
on
al
P
DRB
as
the
ref
e
ren
ce
(
s
k).
This classi
ficat
ion
is
d
e
note
d wit
h
si <s
and
sk
is>
s
k.
Po
te
ntial
an
d
de
velo
ping
sect
or
is
in
(
Q
uadr
ant
III.
T
his
quad
ra
nt
is
a
quadr
a
nt
of
a
sp
e
ci
fic
sect
or
grow
t
h
rate
in
GRDP
(si)
w
hi
ch
is
gr
eat
er
th
an
the
sect
or
gro
w
th
rate
in
t
he
re
gional
G
RDP
as
t
he
re
f
eren
ce
(s)
but
has
a
s
m
al
le
r
sect
or
con
t
rib
ution
value
to
GR
DP
th
an
the
sect
or
c
on
t
rib
ution
to
r
egio
nal
GRD
P
as
the
ref
e
ren
ce
(
s
k).
This classi
ficat
ion
is
d
e
note
d wit
h
si>
s
and
sk
is <s
k.
Unde
rd
e
velo
pe
d
sect
or
is
i
n
Q
ua
drant
IV.
This
qu
a
dr
a
nt
is
a
quad
ra
nt
of
a
sp
e
ci
fic
s
ect
or
gro
wth
rate
in
GRD
P
(si)
w
hich
is
s
m
al
le
r
than
the
sect
or
gro
wth
rate
in
the
re
gio
nal
P
DRB
as
the
ref
e
ren
ce
(
s)
a
nd
al
so
h
as
sm
al
ler
sect
or
co
ntributi
on v
al
ue
t
o
GRDP
(
sk
i)
th
an
t
he
sect
or
c
on
t
ri
buti
on
to
r
egio
nal G
RDP
as
th
e
ref
e
ren
ce
(
s
k).
This classi
ficat
ion
is
d
e
note
d wit
h
si <s
and
sk
i <s
k.
5.
RESEA
R
CH MET
HO
DOL
OGY
The
sam
ple
of
this
stud
y
is
Papu
a
P
r
ov
i
nce
in
the
east
ern
par
t
of
I
ndone
sia
.
The
stu
dy
beg
i
ns
wit
h
the
data
colle
ct
ion
of
the
pr
ov
i
ncial
GRD
P
sect
or
data.
Sect
or
data
use
d
are
20
14
a
nd
2015
data
for
29
reg
e
ncies
i
n
P
apu
a
P
rovince
.
The
ne
xt
ste
p
is
to
di
vid
e
t
he
data
i
nto
dat
a
trai
ning
an
d
te
sti
ng
.
The
da
ta
of
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2502
-
4752
Ind
on
esi
a
n
J
E
le
c Eng &
Co
m
p
Sci,
Vo
l.
12
, N
o.
3
,
Dece
m
ber
2
01
8
:
103
7
–
10
44
1040
2014
a
re
us
e
d
as
data
t
rainin
g
w
hile
data
of
2015
a
re
us
e
d
as
data
te
sti
ng.
Furthe
rm
or
e,
bo
t
h
ty
pes
of
da
ta
are
cl
assifi
ed
ba
se
d
on
Klasse
n
ty
po
lo
gy
to
obta
in
the
i
niti
al
cl
assifi
cat
ion
of
t
he
m
ai
ns
ta
y
econom
ic
reg
io
n.
The
ne
xt
ste
p
is
to
est
ablish
the
ba
sic
rul
es
us
in
g
deci
sion
tree
te
ch
niques
to
ob
ta
in
decisi
on
tre
e
as
a
cl
ass
ific
at
ion
too
l
for
the
ne
xt
m
ai
ns
ta
y
econ
om
ic
reg
io
n.
Tw
o
decisi
on
t
ree
te
ch
niques
us
e
d
in
this
st
udy
are
NBTree
an
d
J
48.
Decisi
on
tree
form
ed
is
te
ste
d
to
the
data
te
sti
ng
as
well
as
to
see
the
accuracy
of
th
e
cl
assifi
cat
ion
of
the
m
ai
ns
ta
y
reg
i
on
s
us
in
g
the
decisi
on
tree
m
od
el
.
T
he
decisi
on
tre
e
te
ch
nique
with
t
he
highest l
evel
of acc
ur
acy
is
use
d
as
the
f
oundat
ion o
f
t
he
m
ai
n
r
ule in
this
stud
y.
6.
PROP
OSE
D MO
DEL OF
THE
MAI
NST
AY
E
CONO
MIC
RE
GIO
N CLAS
SIFI
CA
TI
ON
This
stu
dy
devel
op
s
a
m
od
el
of
t
he
m
ai
ns
ta
y
ec
onom
ic
reg
io
n
cl
assifi
cat
ion
base
d
on
GRDP
sect
or
data
owne
d
by
a
reg
io
n.
Fig
ur
e
1
shows
t
he
de
velo
pe
d
m
od
el
.
The
de
velo
ped
m
od
el
is
a
co
m
bin
at
ion
of
decisi
on
tree
and
Klassen
ty
polo
gy
te
chn
iq
ue
s
as
the
basis
for
determ
ining
the
cl
assifi
cat
ion
of
the
m
ain
s
ta
y
econom
ic
reg
ion.
GR
DP
sect
or
data
of
a
re
gion
in
t
he
pe
r
iod
of
previ
ous
n
ye
ars
a
re
use
d
to
f
or
m
decisi
on
tree
us
i
ng
deci
sion
t
ree.
GR
DP
sect
or
data
are
the
n
cl
ass
ifie
d
usi
ng
Kl
assen
ty
polo
gy
to
obta
in
t
he
init
ial
cl
assifi
cat
ion
r
esults.
T
he
res
ult
of
this
init
ia
l
cl
assifi
cat
ion
is
us
e
d
as
da
ta
trai
ning
for
decisi
on
tree
m
aker
s
us
in
g
decisi
on tree.
T
he
ne
xt stage
is
to
te
st t
he
data
te
sti
ng
to
te
st
the
deci
sion
tree
al
rea
dy
form
ed.
The
m
ai
n
ou
t
pu
t
of
the
de
velo
ped
m
od
el
is
the
cl
assifi
cat
ion
of
t
he
m
ai
n
sta
y
eco
nom
ic
reg
io
n
ba
s
ed
on
t
he
value
of
th
e
GRDP sect
or
da
ta
o
w
ne
d by a
r
e
gion.
Figure
1. Pro
pose
d
Mo
del
of
Ma
instay
Econ
om
ic
Region
C
la
ssific
at
ion
7.
RESU
LT
S
A
ND
DI
SCUS
S
ION
The
i
niti
al
ph
a
se
of
this
stu
dy
was
c
onduct
ed
by
cl
assify
ing
29
reg
e
ncie
s
in
Pa
pua
P
r
ovince
us
i
ng
Klassen
.
T
he
m
ai
n
obj
ect
ive
is
to
m
ake
it
as
data
trai
ni
ng
on
the
f
or
m
at
ion
of
cl
as
sific
at
ion
r
ules
us
in
g
decisi
on
tree
te
chn
i
qu
e
s
of
N
BTree
an
d
J
48.
Table
2
s
hows
the
res
ults
of
t
he
cl
assifi
cat
io
n
of
29
re
ge
nci
es
in
Papua P
r
ov
i
nc
e b
ase
d o
n Klassen.
The
res
ult
of
c
la
ssific
at
ion
at
early
sta
ge
of
this
stud
y
sho
w
s
that
there
are
three
(10.3
4%
)
reg
e
ncies
are
cat
eg
ori
zed
into
t
he
first quadra
nt
(
K1)
i.
e.
the r
e
gio
n
w
it
h
ad
van
c
ed
de
velo
pm
ent
le
vel
an
d
ra
pid
growth
.
The
re
gions
c
at
egorized
int
o
K
1
li
kely
serv
e
as
t
he
m
ai
ns
ta
y
econo
m
ic
reg
ion
a
r
e
Jay
apura
an
d
Pa
niai
Re
gen
ci
es
an
d
Jay
apu
ra
Mu
nicipal
it
y.
It
c
an
be
seen
f
r
om
the
cl
assifi
cat
ion
resu
lt
s
sh
owin
g
that
three
reg
e
ncies
are
cat
egorized
in
to
reg
i
on
s
with
ad
van
ce
d
de
velo
pm
ent
and
ra
pid
gro
wt
h
(
K1).
24.
13
%
of
reg
e
ncies
in
P
apu
a
P
r
ov
i
nce
are
cl
assifi
ed
into
ad
van
ce
d
but
de
pr
esse
d
re
gions,
31.
03%
of
reg
e
nc
ie
s
are
Evaluation Warning : The document was created with Spire.PDF for Python.
Ind
on
esi
a
n
J
E
le
c Eng &
Co
m
p
Sci
IS
S
N:
25
02
-
4752
Cl
as
sif
ic
ation
of
the
Mai
ns
ta
y Eco
nomic R
e
gion
Using
De
ci
sion
Tree Me
thod
(
Her
u Is
m
an
t
o
)
1
041
cat
egorized
i
nto
po
te
ntial
an
d
dev
el
op
i
ng
r
egio
ns
,
w
hi
le
the
rem
ai
nin
g
34.48%
a
re
re
gions
with
r
el
at
ively
unde
rd
e
velo
pe
d
de
velo
pm
ent
sta
tus.
In
the
nex
t
sta
ge,
the
res
ults
of
this
cl
assifi
cat
ion
ar
e
then
use
d
a
s
trai
ning
data
f
or
the
f
or
m
at
i
on
of
decisi
on
tree
us
ing
de
c
isi
on
tree.
As
aforem
entione
d
befo
re,
tw
o
decisi
on
tree
te
ch
niques
us
e
d
i
n
t
his
st
ud
y
are
the
N
BTree
a
nd
J
48
al
gorithm
s.
I
n
this
st
ud
y,
W
eka
t
oo
l
is
use
d
for
decisi
on tree
for
m
at
ion
proc
e
ss.
Table
2.
Cl
assi
ficat
ion
of D
e
ve
lop
m
ent Q
ua
drant i
n 29 Re
ge
ncies in
Pa
pu
a
Pr
ovi
nce
No
District
GRDP
2014
GRDP
2015
Qu
ad
rant
1
As
m
at
7
8
8
.328,6
1
8
3
1
.082,4
9
K4
2
Biak
2
.15
8
.964,4
9
2
.25
4
.816,9
2
K2
3
Bo
v
en
Digu
l
2
.34
6
.150,9
6
2
.46
8
.482,7
4
K2
4
Deiy
ai
4
2
5
.336,8
8
4
7
1
.671,6
0
K3
5
Do
g
iy
ai
3
6
6
.619,9
8
3
9
2
.533,3
7
K3
6
Intan
Jay
a
4
1
2
.149,9
8
4
5
2
.116,7
7
K3
7
Kab
Jayapu
ra
5
.03
8
.190,9
7
5
.55
7
.746,9
5
K1
8
Jayawija
y
a
2
.41
6
.172,1
1
2
.57
8
.258,7
6
K2
9
Kero
m
1
.22
4
.239,7
0
1
.30
8
.614,7
0
K4
10
Lanyja
y
a
5
4
7
.523,9
0
5
8
0
.163,3
6
K4
11
Me
m
b
e
ra
m
o
Ray
a
4
4
0
.824,5
3
4
7
6
.822,5
2
K3
12
Me
m
b
e
ra
m
o
T
en
g
ah
3
4
4
.236,3
0
3
6
6
.598,5
8
K4
13
Mapp
i
9
5
3
.121,3
1
1
.01
8
.560,2
1
K4
14
Mer
au
k
e
5
.25
2
.312,3
0
5
.58
6
.617,6
8
K2
15
Mi
m
ik
a
5
1
.01
3
.4
9
7
,45
5
4
.32
6
.8
4
8
,32
K2
16
Nab
ire
4
.14
3
.384,6
3
4
.42
1
.359,0
0
K2
17
Nd
u
n
g
a
3
7
2
.137,8
9
4
0
7
.087,3
5
K3
18
Pan
iai
1
.85
2
.212,2
7
2
.03
3
.474,7
8
K1
19
Peg
u
n
u
n
g
an
Bin
ta
n
g
7
0
0
.783,0
9
7
2
3
.898,8
1
K4
20
Pu
n
cak J
ay
a
5
5
4
.683,9
2
5
9
5
.277,1
2
K3
21
Pu
n
cak
3
8
1
.722,8
6
4
1
2
.594,9
3
K3
22
Sar
m
i
9
9
1
.923,8
3
1
.05
7
.063,7
6
K4
23
Su
p
riori
4
0
4
.556,8
2
4
1
7
.100,9
7
K4
24
Tolik
ara
5
0
4
.607,8
5
5
2
9
.156,5
9
K4
25
W
arop
en
2
4
4
,60
3
2
8
,30
K3
26
Yah
o
k
i
m
o
6
5
0
.159,2
2
6
9
0
.497,4
3
K4
27
Yali
m
o
3
4
7
.173,1
5
3
7
8
.228,0
6
K3
28
Yap
en
1
.61
5
.976,2
0
1
.70
8
.539,1
0
K2
29
Ko
ta Jay
ap
u
ra
9
.43
4
.791,4
0
1
0
.25
1
.8
6
3
,96
K1
In
al
gorithm
t
est
ing
,
GR
DP
sect
or
data
both
in
2014
a
nd
2015
are
us
ed
as
cl
ass
de
te
rm
inant
of
cl
assifi
cat
ion
r
esult.
T
her
e
ar
e
18
at
trib
utes
us
e
d,
nam
el
y:
ag
ricult
ure,
li
vestoc
k,
f
or
est
ry,
fis
her
y
(
2014_S1
and
2015_S1
)
;
m
ining
and
extracti
on
se
ct
or
s
(
2014
_
S
2
an
d
20
15_S2)
;
m
anu
factu
rin
g
industry
sect
ors
(20
14_S3
a
nd
2015
_S3)
;
el
ect
rici
ty
,
gas
and
wate
r
sec
tors
(
2014_S4
an
d
2015
_S4
);
co
ns
tr
uctio
n
sect
or
(20
14_S5
a
nd
2015_S5
)
;
tra
de,
ho
te
ls
a
nd
restau
ran
ts
sec
tors
(20
14_S6
and
2015
_S6)
;
trans
portat
ion
an
d
com
m
un
ic
at
ion
sect
ors
(20
14_S7
a
nd
2015_S7
);
fina
nce
,
real
est
at
e
a
nd
c
orp
or
at
e
s
erv
ic
es
(
2014
_S8
an
d
2015_S8
); a
nd
serv
ic
e
sect
ors
(20
14_S9 a
nd
2015_S9
).
Test
res
ults
of b
ot
h
al
go
rithm
s
sho
w
that
J
48
al
go
rithm
has
a
bette
r
accu
r
acy
than
NBT
r
ee
,
s
hown
in
Table
3
.
F
r
om
29
data
i
ns
ta
nc
es
te
ste
d,
19
data
are
cat
e
gorize
d
int
o
inc
orrectl
y
cl
assifi
ed
insta
nce,
t
hu
s
the
inaccu
rate
dec
isi
on
tre
e
f
or
m
ed
us
es
NB
TRe
e
of
65.57%.
Me
an
wh
il
e,
f
or
J
48
al
gorithm
,
the
in
accurat
e
decisi
on
tree
f
or
m
ed
is
s
m
all
er
i.e.
31.
03
%
.
Table
3
show
s
the
com
par
ison
of
decisi
on
tree
form
ation
us
i
ng
NBTree
a
nd
J
48
al
gorithm
se
en
from
the
value
of
cl
assifi
c
at
ion
accu
racy,
Kappa
value
,
m
ean
abso
l
ute
error,
and r
oot m
ean sq
ua
re e
rro
r.
Table
3.
C
om
par
iso
n of NBT
r
ee an
d
J
48 Tes
ti
ng
Res
ults
Alg
o
rith
m
s
J4
8
NB
-
Tr
ee
Clas
sif
icatio
n
acc
u
racy (
%
)
6
8
.96
3
4
.48
Kap
p
a
0
.55
5
0
.15
5
Mean abs
o
lu
te E
r
r
o
r
0
.22
9
0
.31
3
Ro
o
t
m
ean s
q
u
are
d
er
ror
0
.37
2
0
.42
1
The
form
at
ion
of
decisi
on
t
re
e
us
i
ng
J48
al
gorithm
sh
ows
t
hat
the
cl
assi
ficat
ion
of
m
ai
n
sta
y
econom
ic
reg
i
on
is
m
or
e
influ
e
nce
d
by
at
tribu
te
of
el
ect
rici
ty
,
gas
and
water
se
ct
or
s
f
or
data
of
2014
(
2014
_S4)
.
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2502
-
4752
Ind
on
esi
a
n
J
E
le
c Eng &
Co
m
p
Sci,
Vo
l.
12
, N
o.
3
,
Dece
m
ber
2
01
8
:
103
7
–
10
44
1042
Figures
2a
a
nd
2b
res
pecti
vel
y
sh
ow
t
he
dec
isi
on
tree
f
or
m
ed
a
nd
the
ru
le
s
ge
ner
at
e
d
f
r
om
the
decisi
on
tree
form
ation
proc
ess u
si
ng J
48.
(a)
(b)
Figure
2. (a
) D
eci
sion
T
ree
of
J48
Re
s
ults (
b) De
ci
sion t
ree
form
ation
proc
ess u
si
ng
J
48
The
ne
xt
sta
ge
is
te
sti
ng
the
data
te
sti
ng
(20
15
GRDP
sect
or
data)
in
to
the
decisi
on
tree
that
is
form
ed.
Af
te
r
wards,
t
he
res
ul
t
of
2015
GR
DP
data
cl
assif
ic
at
ion
us
i
ng
de
ci
sion
tree
in
Figure
2a
is
co
m
par
ed
with
the
res
ult
of
cl
assifi
cat
ion
us
in
g
Klass
en
w
hich
ha
s
been
c
onduct
e
d
earli
er.
Am
on
g
the
29
reg
e
ncies
in
Papua
Prov
i
nc
e
based
on
cl
a
ssific
at
ion
us
i
ng
decisi
on
tree
to
the
data
of
2015,
the
re
are
three
reg
i
ons
wh
ic
h
are
in
dicat
ed
a
s
m
ai
ns
ta
y
reg
ion
s
.
In
this
c
as
e,
the
res
ult
of
J4
8
decisi
on
tr
ee
has
di
ff
e
rence
s
espe
ci
al
ly
from
the r
e
gion ide
nt
ifie
d
as a m
a
i
ns
ta
y ec
onom
i
c reg
i
on. Th
e
r
esult of
re
gion
al
classi
ficat
ion
u
si
ng
J48 s
hows
t
hat
there
is
no
reg
i
on
w
hich
is
cat
egorized
i
nto
K1
or
re
gion
with
ad
va
nced
dev
el
op
m
ent
le
vel
an
d
ra
pid
grow
t
h.
Most
reg
e
n
ci
e
s
in
Papua
Prov
i
nce,
based
on
the
J
48
cl
assifi
cat
ion
,
a
re
cat
ego
rize
d
f
al
l
K2
,
K
3,
a
nd
K
4.
The
m
easur
em
ent
of
acc
ur
ac
y
us
ing
m
eans
sq
ua
re
er
ror
to
the
Klassen
re
su
lt
cl
assifi
cat
i
on
a
nd
J
48
De
ci
sio
n
Tree
s
how
tha
t
the
accu
racy
le
vel
is
65.
51
%.
Ta
ble
4
s
hows
the
com
par
iso
n
of
cl
ass
ific
at
ion
resu
lt
s
us
i
ng
Klassen
and
J48
Decisi
on T
re
e.
Table
4.
C
om
par
iso
n of Klass
en
a
nd J
48 Cl
assifi
cat
ion
s
No
District
GRDP
2015
Klass
en
J4
8
1
As
m
at
8
3
1
.082,4
9
K4
K4
2
Biak
2
.25
4
.816,9
2
K2
K2
3
Bo
v
en
Digu
l
2
.46
8
.482,7
4
K2
K4
4
Deiy
ai
4
7
1
.671,6
0
K3
K3
5
Do
g
iy
ai
3
9
2
.533,3
7
K3
K3
6
Intan
Jay
a
4
5
2
.116,7
7
K3
K3
7
Kab
Jay
ap
u
ra
5
.55
7
.746,9
5
K1
K2
8
Jayawija
y
a
2
.57
8
.258,7
6
K2
K2
9
Kero
m
1
.30
8
.614,7
0
K4
K2
10
Lanyja
y
a
5
8
0
.163,3
6
K4
K3
11
Me
m
b
e
ra
m
o
Ray
a
4
7
6
.822,5
2
K3
K3
12
Me
m
b
e
ra
m
o
T
en
g
ah
3
6
6
.598,5
8
K4
K3
13
Mapp
i
1
.01
8
.560,2
1
K4
K4
14
Mer
au
k
e
5
.58
6
.617,6
8
K2
K2
15
Mi
m
ik
a
5
4
.32
6
.8
4
8
,32
K2
K2
16
Nab
ire
4
.42
1
.359,0
0
K2
K2
17
Nd
u
n
g
a
4
0
7
.087,3
5
K3
K2
18
Pan
iai
2
.03
3
.474,7
8
K1
K4
19
Peg
u
n
u
n
g
an
Bin
ta
n
g
7
2
3
.898,8
1
K4
K4
20
Pu
n
cak J
ay
a
5
9
5
.277,1
2
K3
K4
21
Pu
n
cak
4
1
2
.594,9
3
K3
K3
22
Sar
m
i
1
.05
7
.063,7
6
K4
K4
23
Su
p
riori
4
1
7
.100,9
7
K4
K4
24
Tolik
ara
5
2
9
.156,5
9
K4
K4
25
W
arop
en
3
2
8
,30
K3
K4
26
Yah
o
k
i
m
o
6
9
0
.497,4
3
K4
K4
27
Yali
m
o
3
7
8
.228,0
6
K3
K3
28
Yap
en
1
.70
8
.539,1
0
K2
K2
29
Ko
ta Jay
ap
u
ra
1
0
.25
1
.8
6
3
,96
K1
K2
2014_S4 <
=
486
|
2014_S4
<=
67.
049972:
K3
(
10.
0/2.
0)
|
2014_S4
>
6
7.
049972:
K4
(1
1.
0/3.
0)
2014_S4 >
486:
K2 (8.
0/2
.
0)
Evaluation Warning : The document was created with Spire.PDF for Python.
Ind
on
esi
a
n
J
E
le
c Eng &
Co
m
p
Sci
IS
S
N:
25
02
-
4752
Cl
as
sif
ic
ation
of
the
Mai
ns
ta
y Eco
nomic R
e
gion
Using
De
ci
sion
Tree Me
thod
(
Her
u Is
m
an
t
o
)
1043
In
Table
4,
th
ere
are
10
re
ge
ncies
with
diff
e
ren
t
cl
assifi
cat
ion
resu
lt
s
betwee
n
Klass
en
a
nd
J
48.
The
c
ha
ng
e
s
of
Klassen
cl
as
sific
at
ion
posit
ion
occ
urre
d
i
n
10
re
ge
ncies
,
nam
el
y:
Bov
en
Digul
Re
ge
ncy
is
cat
egorized
int
o
K
2
in
Klasse
n
but
K4
in
J
48;
Jay
apura
Re
gen
cy
f
ro
m
K1
to
K2
;
Kerom
Re
gen
cy
f
ro
m
K4
to
K2
;
La
nyj
ay
a Regency from
K4
t
o
K3
;
Me
m
ber
a
m
o
Tengah
Re
ge
ncy
f
rom
K4
to K3;
N
dunga
Re
ge
nc
y
fr
om
K3
t
o
K2
;
Pa
ni
ai
Re
gen
cy
f
r
om
K1
to
K
4;
P
un
ca
k
Jay
a
Re
gen
cy
from
K3
to
K
4;
W
a
r
open
Re
ge
ncy
f
rom
K3
to
K
4;
an
d
Kota
Jay
apu
r
a
f
rom
K1
to
K
2.
T
her
e
fore,
base
d
on
th
e
res
ults
of
J48
cl
assif
ic
at
ion
,
the
re
a
re
10
reg
e
ncies
wh
ic
h
a
re
cat
eg
ori
zed
int
o
the
m
ain
sta
y
eco
nom
i
c
re
gion
with
t
he
le
vel
of
a
dva
nced
but
depr
essed
dev
el
op
m
ent (K2
)
as
see
n
in
Table
5.
Table
5.
Re
ge
nc
ie
s w
it
h
Ma
in
sta
y Econom
ic
Re
gion
No
District
GRDP
2015
Klass
en
J4
8
1
Biak
2
.25
4
.816,9
2
K2
K2
2
Kab
Jay
ap
u
ra
5
.55
7
.746,9
5
K1
K2
3
Jayawija
y
a
2
.57
8
.258,7
6
K2
K2
4
Kero
m
1
.30
8
.614,7
0
K4
K2
5
Mer
au
k
e
5
.58
6
.617,6
8
K2
K2
6
Mi
m
ik
a
5
4
.32
6
.8
4
8
,32
K2
K2
7
Nab
ire
4
.42
1
.359,0
0
K2
K2
8
Nd
u
n
g
a
4
0
7
.087,3
5
K3
K2
9
Yap
en
1
.70
8
.539,1
0
K2
K2
10
Ko
ta Jay
ap
u
ra
1
0
.25
1
.8
6
3
,96
K1
K2
Table 5 sh
ows
te
n
re
gen
ci
es
, nam
ely Bi
ak,
Jaya
pura,
Jay
a
wij
ay
a,
Kerom
, Merauke,
Mi
m
ika, N
abi
re,
Ndu
ng
a
, Yape
n
a
nd Jaya
pura
Mun
ic
i
palit
y.
8.
CONCL
US
I
O
N
Ba
sed
on
t
he
r
esults
of
the
stud
y,
decisi
on
t
ree
te
ch
niques
can
be
use
d
a
s
an
al
te
r
native
appr
oach
t
o
determ
ine
the
m
ai
ns
ta
y
econ
om
ic
reg
ion
.
T
he
res
ults
sho
w
that
both
Kl
assen
an
d
J
48
decisi
on
trees
ind
ic
at
e
d
that
Jay
apu
ra
Mun
ic
ipali
ty
and
Jay
ap
ura
Re
gen
cy
are
sti
ll
the
m
ai
ns
ta
y
econom
ic
reg
ion
s,
al
th
ough
ba
sed
on
reg
i
on
al
cl
assif
ic
at
ion
resu
lt
s,
bo
th
are
cat
e
gorize
d
into
dif
f
eren
t
cl
ass
when
they
are
cl
assifi
ed
with
Kl
assen
and
J48.
I
n
a
ddit
ion
,
the
acc
ur
acy
le
vel
of
2015
GR
DP
se
ct
or
data
te
sti
ng
to
t
he
decisi
on
tree
J
48
sho
ws
that
the
accu
racy
is
65.51%.
The
resu
lt
s
of
Klassen
s
how
that
there
are
t
hr
ee
reg
e
ncies
that
are
cat
eg
or
iz
e
d
int
o
the
m
ai
ns
ta
y
e
conom
ic
reg
io
n.
Me
a
nwhile
,
the
res
ults
of
decisi
on
tree
J
48
sho
w
that
t
her
e
a
re
10
re
gen
ci
es
that
are
cat
eg
ori
zed
int
o
the
m
ai
ns
ta
y
econom
ic
reg
ion
.
Ther
e
f
or
e,
dec
isi
on
tree
te
c
hniq
ue,
es
pecial
ly
J4
8
al
gorithm
, can
b
e use
d
as a
n
a
lt
ern
at
ive in cl
assify
ing
reg
i
ons
int
o
certai
n m
ai
ns
ta
y regions. As a r
e
su
lt
,
it
can
be use
d
as
poli
cy
m
aking
m
at
erial
s for
l
ocal
gove
rn
m
ents to det
erm
ine the m
a
instay
eco
no
m
ic
r
egio
n.
REFERE
NCE
S
[1]
Karsina
h,
Putri
P.I.
,
Rah
a
y
u
K.
N,
Panji
putri
A.
F.
The
Profile
of
Pekal
ongan
a
s
a
Cent
er
of
Econom
ic
Grow
th
at
Ta
ngka
ll
angk
a
Strat
egic
Are
as.
Inte
rnationa
l
J
ournal
of
Ec
ono
mics
and
Fi
nan
ci
al
Iss
ues
,
201
6;
Vol
6,
Spe
cia
l
Iss
ue
(S6).
[2]
Munanda
r
T.
A.
,
Azha
ri,
Mus
dholi
fah
A.,
Ars
y
ad
L.
Hierarchi
cal
Re
g
ional
Dispar
it
ie
s
and
Pot
ential
Se
ct
or
Ide
nti
f
i
ca
ti
on
Us
ing
Modif
ie
d
Aggl
omerati
ve
Cluste
ring
.
Proc
ee
ding
of
Th
e
1st
Annual
Applie
d
Sci
enc
e
an
d
Engi
ne
eri
ng
Con
fer
ence
(AA
SEC
2016).
[3]
Ramani
R.
G
,
Shanthi
S.
Classi
fi
er
Pred
ic
t
ion
Ev
aluation
in
M
odel
ing
Road
Tr
aff
ic
Accid
ent
D
ata,
Conf
ere
nc
e
Proce
edi
ng
of
C
om
puta
ti
ona
l
Int
el
li
g
ence &
Co
m
puti
ng
Resea
r
c
h
(ICCIC).
2012
;
Page(
s):
1
–
4.
[4]
Kam
ber
,
Jiawe
i H
,
Mi
che
l
ine.
Data
Min
ing:
Co
nce
pt
and
Techn
ique
s.
CA
US
A.
Aca
demi
c
Press
.
2001
.
[5]
Nisw
at
in
R.
K,
Sanjay
a
A.
Cl
assific
a
ti
on
Of
Cat
egor
y
S
el
e
ction
T
it
l
e
Under
gra
duate
The
sis
Us
ing
K
-
Nea
re
st
Neighbor
Metho
d.
2017;
Indone
sian
Journal
of
El
e
ct
rica
l
Engi
n
ee
ring
and
Computer
Sci
en
ce.
Vol.
7.
No.
3,
pp
.
846
~ 854.
[6]
Bresfe
lean
V.
P.
,
Bresfe
le
an
M.,
Ghisoiu
N.
Det
ermining
Stude
n
ts’
Ac
ademi
c
Fa
il
ure
Prof
il
e
Fo
unded
on
Dat
a
Mini
ng
Me
thods
.
Confer
ence
Proce
ed
ing
of
30th
Inte
rna
t
iona
l
Confer
ence
on
Inform
at
ion
Te
ch
nolog
y
Int
erf
a
ces
.
2008;
Page(
s):
3
17
–
322
.
[7]
Hus
sain
S,
Dah
an
N.A,
Ba
-
Alwi
F.M,
Riba
ta
N.
Educ
ationa
l
Data
Mining
an
d
Anal
y
sis
of
Student
s’
Aca
demi
c
Perform
anc
e
Us
i
ng
W
EKA.
Ind
onesian
Journal
of
E
lectric
a
l
E
ngine
ering
and
Computer
Scien
ce
.
2018;
Vol
.
9
,
No.
2.
Februa
r
y
.
pp
.
447~459
[8]
Slaught
er
G,
K
urtz
Z,
d
esJardi
ns
M,
Hu,
P.
F,
Mac
k
enzie
C
,
Stansbur
y
FR
CA
L,
Stei
n
D
M,
R
ed
ic
t
ion
of
Mortali
ty
.
Proc
e
edi
ng
of
IE
EE Bi
o
m
edi
cal
Cir
cuits a
nd
S
y
stems
Confer
ence
(B
io
CAS
).
2012;
Pag
e(s)
1
–
4.
[9]
Munanda
r
T.
A,
W
ina
rko
E.
Regi
onal
Dev
el
opm
ent
Cla
ss
i
fic
a
ti
on
Model
using
Dec
ision
Tre
e
Approa
ch.
Inte
rnational
Jo
urnal
of
Comput
er
Applications
,
2015;
Volum
e
1
14
–
No.
8,
pp.
2
8
–
33
.
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2502
-
4752
Ind
on
esi
a
n
J
E
le
c Eng &
Co
m
p
Sci,
Vo
l.
12
, N
o.
3
,
Dece
m
ber
2
01
8
:
103
7
–
10
44
1044
[10]
Mori,
Hiro
y
uk
i.
Appl
ic
a
ti
on
of
NBTr
ee
to
Selec
t
ion
of
Me
t
eo
rologic
al
Varia
ble
s
in
Wind
S
pee
d
Pred
ic
t
ion
.
Confer
ence
Proc
ee
ding
of
Tr
ansm
ission
&
Distri
buti
on
Confer
en
ce
&
Exposition
:
As
ia
and
Paci
fi
c.
2009;
Page(
s)
:
1
–
4.
[11]
Kohavi
R,
Sca
ling Up t
he
A
cc
ur
acy
o
f
Naï
ve
-
Ba
ye
s Classif
ie
rs
: a Dec
ision
-
Tr
ee
Hybrid
.
AA
AI. 1996.
[12]
Zha
o
Y,
Zha
ng
Y.
Com
par
ison
of
dec
ision
tree
m
et
hods
for
findi
ng
ac
t
ive
objects
.
Adv
an
ce
s
in
Space
Re
search
.
2008;
41(12)
,
19
55
–
1959.
[13]
Ga
y
at
ri
N,
Ni
ck
ola
s
S,
R
edd
y
A.V,
Chit
r
a
R
.
Pe
rform
ance
An
aly
sis
Of
Data
Mini
ng
Al
gorith
ms
For
Soft
ware
Qualit
y
Pred
ic
t
i
on
.
Confer
en
ce
Proce
edi
ng
of
I
nte
rna
ti
ona
l
Confer
ence
on
Advanc
es
in
Recen
t
Te
chno
logi
es
i
n
Com
m
unic
at
ion and
Com
puti
ng.
2009;
Page(
s)
:
3
93
–
395
.
[14]
Go
y
al
A,
Meh
t
a
R.
Pe
rform
anc
e
compari
son
of
Naïve
B
a
y
es
and
J48
c
la
ss
if
ic
a
ti
on
al
gori
th
m
s
.
Inte
rnation
al
Journal
of
Appli
ed
Eng
ine
ering
Re
search
,
2012;
7
(11
SU
PP
L.
),
1389
–
1393.
[15]
As
wandi
H,
Kuncoro
M,
Evalua
si
Penetapa
n
K
a
wasan
Andala
n
:
Studi
Empir
is
d
i
Kal
imanta
n
Se
l
at
an
1993
-
1998.
Jurnal
E
konomi
dan
Bi
snis Indon
esia
.
2002
;
Vol
.
17,
No.
1,
pp.
27
–
45
.
in
Bah
asa
.
[16]
Toda
ro
M.
P.
Econom
ic
Deve
lop
m
ent
,
Sev
ent
h
E
dit
ion. New
Yor
k.
Addit
ion
W
es
ley
Longman.
In
c.
2000
.
[17]
Sutarno,
Kunco
ro
M.
Keti
m
pa
ngan
Antar
K
e
ca
m
at
an
Di
Ka
bupat
en
B
an
y
u
m
as
1993
-
2000.
Jurnal
E
k
ono
mi
Pe
mbangunan
,
2
000;
8(2). 97
–
11
0.
in
Bah
asa
.
Evaluation Warning : The document was created with Spire.PDF for Python.