Enginee ing and Technology Jou nal e-ISSN: 2456-3358
Volume 10 Issue 11 No embe -2025, Page No.-7916-7922
DOI: 10.47191/e j/ 10i11.22, I.F. – 8.482
© 2025, ETJ
7916
ETJ Volume 10 Issue 11 No embe 2025,
1
M. Tau ik Ap inaldo
Clus e ing o Sc ew P ess Machine Condi ions using he K-Medoids Me hod
M. Tau ik Ap inaldo1, Jas il2*, Suwan o Sanjaya3, Les a i Handayani4, Fi i Insani5
1,2,3,4,5In o ma ics Enginee ing, Facul y o Science and Technology, Sul an Sya i Kasim Riau S a e Islamic Uni e si y.
ABSTRACT: The sc ew p ess plays an impo an ole in he oil ex ac ion p ocess; hus, moni o ing i s condi ion is essen ial o
main ain pe o mance and p e en ailu es. This s udy aims o clus e sc ew p ess machine condi ions using he K-Medoids me hod.
The da ase consis ed o 23,002 eco ds om PT. XYZ was collec ed in Ap il–May 2024 wi h wo a ibu es: empe a u e and
p essu e. The da a was p ocessed h ough selec ion, p e-p ocessing, and ans o ma ion s ages using z-sco e no maliza ion be o e
clus e ing. Model e alua ion employed he Silhoue e Coe icien and he Da ies-Bouldin Index (DBI). The esul s show ha he
bes con igu a ion was a K = 7, wi h a Silhoue e alue o 0.5494 and a DBI o 0.5521, indica ing a easonable s uc u e and good
sepa a ion. Thus, he K-Medoids me hod has been p o en e ec i e in clus e ing sc ew p ess machine condi ions and use ul in
suppo ing machine main enance decision-making.
KEYWORDS: Clus e ing, K-Medoids, Silhoue e Coe icien , Da ies-Bouldin Index, Sc ew P ess.
I. INTRODUCTION
The sc ew p ess machine is a ool ha plays an impo an
ole in he oil ex ac ion p ocess, whe e he selec ion o
app op ia e pa ame e s is necessa y o main ain he quali y o
aluable subs ances in he ex ac ed oil [1]. In manu ac u ing
indus ies such as palm oil p ocessing, his machine has
se e al impo an componen s, such as wo m sc ews,
ex ension sha s, bea ings, p ess cages, and oil seals ha
equi e ou ine main enance [2]. The pe o mance o he
p oduc ion machine i sel is highly dependen on he le el o
eliabili y and a ailabili y, whe e damage (down ime) o he
sc ew p ess o en causes he p oduc ion p ocess o be
subop imal and he company's a ge s o no be achie ed [3].
Addi ionally, esea ch on sc ew p ess hub damage ound ha
ic ion and dynamic loads igge ma e ial a igue, ini ial
c acks, and e en b eakage be o e ou mon hs o use [4].
Gi en he ope a ional complexi y o sc ew p ess
machines and he impo ance o op imal pa ame e s in hei
pe o mance, a sys ema ic app oach is needed o moni o and
analyze he condi ion o hese machines. Clus e ing is an
unsupe ised da a analysis echnique used o g oup da a
based on speci ic pa e ns and simila i ies, which plays an
impo an ole in a ious ields, including machine condi ion
moni o ing [5]. This echnique becomes pa icula ly ele an
when combined wi h Machine Lea ning (ML) echnology o
analyze machine ope a ional da a in g ea e dep h. In i s
indus ial applica ion, ML can be used as a p edic i e
main enance me hod by u ilizing da a collec ed om IoT
de ices ins alled on machines o de ec ea ly aul s and
p e en majo ailu es, as demons a ed in he case o kni ing
machines wi h an accu acy a e o 92% [6]. Speci ically,
clus e ing helps in analyzing uns uc u ed and high-
dimensional da a in he o m o sequences, exp essions,
ex , and images [7].
P e ious esea ch has applied he Fuzzy C-Means
algo i hm o clus e sc ew p ess machine condi ions based on
empe a u e and p essu e. The pe o mance o he sc ew p ess
machine signi ican ly a ec s he quali y and e iciency o
palm oil p oduc ion. The Back P essu e Vessel (BPV), which
is esponsible o dis ibu ing s eam o a ious p ocess
s a ions, is an impo an pa o his sys em [8]. Meanwhile,
he K-Means algo i hm has also been used on he same da ase
wi h es ing o up o i een clus e con igu a ions. Based on
DBI, he bes quali y was ob ained wi h h ee clus e s, while
he Elbow me hod indica ed ou clus e s as he op imal
choice [9]. Bo h s udies con i m ha clus e ing echniques
a e e ec i e in desc ibing machine ope a ional pa e ns,
al hough hey a e s ill limi ed o he use o ce ain a iables,
hus equi ing u he s udy wi h o he algo i hms such as K-
Medoids.
The e o e, he esea che will conduc esea ch on
clus e ing he condi ions o sc ew p ess machines. The main
ocus o his s udy is o clus e sc ew p ess machine
condi ions using he k-medoids me hod. K-medoids is a
pa i ion-based clus e ing algo i hm ha uses ac ual poin s as
medoids, which a e he mos cen al poin s in a clus e [10].
The selec ion o he k-medoids me hod is based on i s
ad an ages in handling machine condi ion da a, whe e, in he
con ex o machine condi ion analysis, K-medoids has been
applied o model unce ain y in la ge-scale elec ical powe
dis ibu ion sys ems, demons a ing high accu acy and
scalabili y compa ed o o he me hods. Fu he mo e, wi h he
applica ion o K- medoids, da a can be o ganized in o mo e
ep esen a i e clus e s, enabling mo e e icien decision-
“Clus e ing o Sc ew P ess Machine Condi ions using he K-Medoids Me hod”
7917
Volume 10 Issue 11 No embe 2025ETJ ,
1
M. Tau ik Ap inaldo
making ela ed o moni o ing and imp o ing sys em
pe o mance [11].
P e ious s udies ha e p o en he e ec i eness o he K-
medoids algo i hm in a ious applica ion domains. This
algo i hm success ully g ouped songs on Spo i y based on
hei popula i y and dis ibu ion in playlis s wi h a Silhoue e
Sco e o 0.5014, which indica es good clus e sepa a ion and
has p ac ical implica ions o he music indus y in designing
mo e a ge ed ma ke ing s a egies [12]. Addi ionally, K-
medoids has also been p o en op imal in classi ying
ea hquake ulne abili y le els in Indonesia wi h k=2 and a
maximum Silhoue e Coe icien o 0.68016, success ully
iden i ying 390 ea hquake e en s wi h a e y high
ulne abili y le el in Eas e n Indonesia and 179 e en s wi h
a high ulne abili y le el in Wes e n Indonesia, p o iding an
impo an e e ence o he go e nmen in planning
ea hquake isk mi iga ion [13]. Fu he mo e, his algo i hm
is also e ec i e in g ouping iolence-p one a eas in he ci y
o Padang using da a om 2,434 cases sp ead ac oss 11
subdis ic s, p oducing 3 clus e s ha can ca ego ize a eas
based on he le el o iolence cases om highes o lowes
[14]. The success o K-medoids in hese h ee s udies
demons a es he consis ency and adap abili y o his
algo i hm in handling a ious ypes o da a and clus e ing
p oblems, making i a s ong ounda ion o i s applica ion in
analyzing he condi ion o sc ew p ess machines.
II. MATERIALS AND METHODS
A. Resea ch S ages
This esea ch ollows a numbe o sys ema ic s eps
designed o analyze and p ocess da a accu a ely. The ini ial
s ep begins wi h selec ing a da ase o de e mine he
in o ma ion o be analyzed. This is ollowed by a p e-
p ocessing s age o clean he da a so ha i is eady o
ans o ma ion. A e he da a is cleaned, he p ocess
con inues o he ans o ma ion phase, which includes da a
no maliza ion. The nex s ep is clus e ing using he k-
medoids algo i hm. The inal s age is e alua ion o de e mine
he op imal numbe o clus e s om he applica ion o he
algo i hm used. Figu e 1 shows he esea ch low.
Figu e 1: Resea ch Flow
B. Da a Selec ion
In his phase, da a cu a ion is ca ied ou o ensu e he
sui abili y o he da ase wi h he esea ch objec i es,
including he selec ion o ele an a iables and he
de e mina ion o da a ypes ha a e in line wi h he
me hodology o be applied.
C. Da a P e-p ocessing
This p e-p ocessing s age is implemen ed o il e and
pu i y he da ase om elemen s ha can hinde he analysis
p ocess. The s ages ca ied ou include emo ing edundan
da a, add essing missing alues, co ec ing w i ing e o s, and
ha monizing inconsis en da a o ma s.
D. Da a T ans o ma ion
In da a ans o ma ion, he e a e a ious commonly used
ans o ma ion echniques such as min-max scaling, z-sco e,
decimal scaling, obus scaling, one-ho encoding, and o he s.
In his s udy, he au ho will use he z-sco e echnique in he
da a ans o ma ion sec ion [15]. The z-sco e Fo mula [16]
can be seen in equa ion (1):
𝑋′=𝑋i−mean(x)
s d(x) (1)
Whe e 𝑋′is he no maliza ion esul alue, 𝑋i is he
no malized da a, mean(x) is he mean alue o an a ibu e,
and s d(x) is he s anda d de ia ion o an a ibu e.
E. K-Medoids Implemen a ion
K-medoids is a da a clus e ing me hod ha can p o ide
mo e balanced esul s in e ms o g oup dis ibu ion
compa ed o o he me hods, such as K-Means [17]. K-
medoids is a pa i ion-based clus e ing algo i hm ha uses
ac ual poin s as medoids, which a e he mos cen al poin s in
a clus e [10]. The s eps in he K-Medoids me hod a e [18] :
a. Ini ialize medoids by andomly selec ing k objec s
om he da ase as ini ial medoids.
b. Nex , each objec is assigned o he clus e wi h he
nea es medoid based on he Euclidean dis ance
using equa ion (2).
𝑑(𝐴,𝐵)=√[(x₁ −x₂)2 + (y₁ −y₂)2] (2)
Explana ion :
𝑑(𝐴,𝐵)
=
Euclidean dis ance be ween
poin A and poin B
𝐴
=
Objec o be calcula ed
𝐵
=
Clus e cen e
x₁
=
The i s -dimensional
coo dina e alue o poin A
x₂
=
The i s -dimensional
coo dina e alue o poin B
y₁
=
The alue o he second
dimension coo dina e o
poin A
y₂
=
The alue o he second
dimension coo dina e o
poin B
c. Then, a new medoid is selec ed whe e, o each
clus e , he o al dis ance om all objec s o o he
objec s in he same clus e is calcula ed.
d. Calcula e he o al de ia ion (S) by calcula ing he
o al new dis ance – o al old dis ance. I S< 0, hen
swap he objec wi h he clus e da a o o m a new
se o k objec s as he medoid. I S> 0, hen he o al
dis ance inc eases and no eplacemen is made (use
“Clus e ing o Sc ew P ess Machine Condi ions using he K-Medoids Me hod”
7918
Volume 10 Issue 11 No embe 2025ETJ ,
1
M. Tau ik Ap inaldo
he old medoid as a e e ence o he nex i e a ion),
using equa ion (3).
𝑆 = ∑Dnew − ∑Dold (3)
Whe e 𝑆 is he o al de ia ion, ∑Dnew is he o al
dis ance o he new medoid, and ∑Dold is he o al
dis ance om he old medoid.
e. Repea s eps C and D un il he e is no change in he
medoid (S=0).
F. Model E alua ion
The e alua ion me hods o be used a e he Silhoue e
Coe icien and Da ies-Bouldin Index (DBI).
a. Silhoue e Coe icien
The Silhoue e Coe icien desc ibes he quali y o a da a
poin 's placemen wi hin i s clus e by compa ing i s
p oximi y o i s own clus e and o he clus e s. This
coe icien anges om -1 o 1. E alua ion using he
silhoue e me hod [19] is calcula ed using equa ion (4).
𝑆(𝑥𝑖)= 𝑏(𝑥𝑖) − 𝑎(𝑥𝑖)
𝑀𝑎𝑥 {𝑏(𝑥𝑖),𝑎(𝑥𝑖)} (4)
Whe e 𝑆(𝑥𝑖) is he silhoue e coe icien alue o he i- h
da a poin , 𝑎(𝑥𝑖) is he a e age dis ance be ween he i- h
objec and o he objec s in he same clus e (in a-clus e ), and
𝑏(𝑥𝑖) is he a e age dis ance be ween he i- h objec and
objec s in he nea es di e en clus e (in e -clus e ).
b. Da ies-Bouldin Index
In o de o maximize he numbe o clus e s o med, he
Da ies-Bouldin Index (DBI) is employed o calcula e how
many clus e s should be o med [20]. Unlike he Silhoue e
Coe icien , a lowe DBI alue indica es be e clus e ing
quali y. DBI has a alue ange om 0 upwa ds, wi h alues
close o 0 indica ing op imal clus e sepa a ion and high
in e nal cohesion. The DBI e alua ion guidelines a e as
ollows: alues less han 1.0 indica e good clus e ing, alues
be ween 1.0 and 2.0 indica e accep able clus e ing, while
alues abo e 2.0 indica e subop imal clus e ing wi h high
o e lap be ween clus e s. The e a e ou s eps in he DBI
e alua ion [21] :
The e alua ion begins wi h calcula ing SSW and SSB using
he ini ial cen oid as a e e ence, ollowed by calcula ing he
DBI alue. The Fo mula applied o ob ain he Sum o Squa e
Wi hin clus e (SSW) is :
𝑆𝑆𝑊 = 1
𝑚∑𝑗=𝑖
𝑚𝑖 𝑑(𝑥𝑗, 𝐶𝑖) (5)
Whe e 𝑚𝑖 is he numbe o da a poin s in clus e i, 𝐶𝑖 is he
cen oid o clus e i, 𝑥𝑗 is he da a in he clus e , and 𝑑(𝑥𝑗, 𝐶𝑖)
is he dis ance be ween he da a and he cen oid.
The Sum o Squa es Be ween-Clus e (SSB) has a unc ion o
analyze he le el o sepa a ion be ween da a g oups. The SSB
me ic se es o e alua e how sepa a ed he clus e s a e by
calcula ing he dis ance be ween he cen e poin o each
clus e (cen oid) and he cen e poin o he en i e da ase .
The highe he SSB alue ob ained, he mo e op imal he
sepa a ion be ween clus e s. The ma hema ical Fo mula
applied in he calcula ion o SSB is :
SSBi,j = d(Ci,Cj) (6)
Whe e d(Ci,Cj) is he dis ance be ween clus e s.
The nex s ep is o calcula e he a io alue o each clus e
using he ollowing Fo mula:
𝑅𝑖,𝑗 =𝑆𝑆𝑊𝑖+𝑆𝑆𝑊𝑗
𝑆𝑆𝐵𝑖,𝑗 (7)
Whe e 𝑅𝑖,𝑗 is he a io o SSW and SSB be ween clus e i and
clus e j, 𝑆𝑆𝑊𝑖 is he Sum o Squa es Wi hin-clus e o
clus e i, 𝑆𝑆𝑊
𝑗 is he Sum o Squa es Wi hin-clus e o
clus e j, and 𝑆𝑆𝐵𝑖,𝑗 is he Sum o Squa es Be ween-clus e
be ween clus e i and j.
The inal phase in he e alua ion p ocess in ol es
calcula ing he DBI by aking he a e age alue o he
maximum a io ound be ween each da a g oup and he o he
da a g oups. The DBI index se es as a pa ame e o measu e
he le el o in e nal compac ness in each g oup and he
quali y o sepa a ion o med be ween da a g oups. The
ma hema ical Fo mula o calcula ing he Da ies-Bouldin
Index can be o mula ed as :
𝐷𝐵𝐼 = 1
𝑘 ∑𝑖
𝑘 𝑚𝑎𝑥(𝑅𝑖,𝑗) (8)
Whe e 𝑅𝑖,𝑗 is he SSW and SSB a io be ween clus e
i and clus e j, and 𝑘 is he numbe o clus e s
o med.
III. RESULT AND DISCUSSION
A. Da a Selec ion
This s udy will p ocess he o iginal da ase om sc ew
p ess de ices wi h BPV de ice iden i ica ion du ing he
pe iod om Ap il o May 2024, ob ained om PT.XYZ, wi h
a o al o 23,002 da a ows. The ini ial da a consis s o no,
ows amp, de ice code, da e, empe a u e, p essu e, pH,
cu en , c ea ed a , c ea ed by, and mo ed a his o y. Then,
da a selec ion is ca ied ou , and wo impo an a iables a e
aken, namely empe a u e and p essu e.
In he nex s age, he main cha ac e is ics o he exis ing
da a we e collec ed and p ocessed o o m a da ase o 23,002
da a poin s used o sys em analysis and modeling. This s udy
applied a ea u e selec ion me hod by aking wo main
pa ame e s, namely Tempe a u e and P essu e, as inpu
a iables o u he da a p ocessing. These wo a ibu es
we e selec ed because hey ha e a signi ican co ela ion wi h
he ope a ional pe o mance and condi ion s a us o he sc ew
p ess de ice. The selec ed da a esul s a e shown in Table 1.
Table 1. Da ase a e da a selec ion
No
Tempe a u e
P essu e
0
123
3,19
1
123
3
...
...
...
23000
100
2,12
23001
100
2,03
B. Da a P e-p ocessing
In he nex s age, impo an a ibu es om he da ase
we e so ed and p ocessed o checking.
“Clus e ing o Sc ew P ess Machine Condi ions using he K-Medoids Me hod”
7919
Volume 10 Issue 11 No embe 2025ETJ ,
1
M. Tau ik Ap inaldo
Table 2. Da a e i ica ion esul s
No
A ibu e
Coun
Null Coun
Da a
Type
1
Tempe a u e
23002
No -Null
Floa 64
2
P essu e
23002
No -Null
Floa 64
A e checking he da a, he esul s show ha he e a e no
null alues in he Tempe a u e and P essu e a ibu es, which
con ain 23002 da a poin s and ha e he same da a ype,
Floa 64, as shown in Table 2.
C. Da a T ans o ma ion
A his s age, da a ans o ma ion is pe o med o
s anda dize he scale be ween a iables wi h di e en alue
anges, namely empe a u e and p essu e, so ha no a iable
is mo e dominan in he dis ance calcula ion p ocess in he K-
Medoids algo i hm. The esul s o he da ase ans o ma ion
in Table 1 using equa ion (1) can be seen in Table 4 below.
Table 3. All da a a e ans o ma ion
No
Tempe a u e
P essu e
0
0,693464
0,826668
1
0,693464
0,588746
...
...
...
23000
0,201872
-0,513208
23001
0,201872
-0,625908
D. K-Medoids Implemen a ion
In his sec ion, based on Table 3, dis ance calcula ions
we e pe o med using equa ion (2) and To al De ia ion using
equa ion (3) un il he e we e no changes in he medoids. The
calcula ion esul s can be seen in Figu e 2 and Table 4 .
Figu e 2: Visualiza ion o clus e ing wi h medoid cen e s
The esul s o applying he K-Medoids me hod o he
en i e da ase can be seen in Figu e 2, which shows he
posi ions o he medoids and he dis ibu ion o clus e s a e
he p ocess eaches a s able condi ion.
Table 4. Dis ibu ion o he numbe o membe s in each
clus e
No
Tempe a u e
P essu e
Clus e
0
0,693464
0,826668
0
1
0,693464
0,588746
0
...
...
...
…
22987
0,201872
0,676402
0
22988
0,693464
0,263169
0
5
0,693464
0,225602
1
6
0,693464
0,213080
1
…
…
…
…
23000
0,201872
-0,513208
1
23001
0,201872
-0,625908
1
743
-1,935483
0,000203
2
744
-1,935483
0,050291
2
…
…
…
…
22996
-1,935483
0,313258
2
22997
-1,935483
0,576224
2
In addi ion, he clus e ing esul s showing he numbe o
membe s in each clus e can be seen in Table 5, namely
clus e 0 con ains 7,356 da a, clus e 1 con ains 11,173 da a,
and clus e 2 con ains 4,473 da a.
E. Model E alua ion
In he e alua ion s age, he clus e ing esul s we e es ed
using wo me hods, namely he Silhoue e Coe icien and he
Da ies-Bouldin Index (DBI). The Silhoue e Coe icien
se es o assess he quali y o sepa a ion be ween clus e s,
while he Da ies-Bouldin Index (DBI) measu es he deg ee
o simila i y be ween clus e s.
a. Silhoue e Coe icien
The i s s ep in he e alua ion p ocess is he applica ion
o he Silhoue e Coe icien me hod, as shown in equa ion
(4). This me hod p o ides a quan i a i e measu e o how well
each da a poin is placed in he app op ia e clus e , aking in o
accoun he compa ison o i s p oximi y o i s own clus e and
i s dis ance om o he clus e s. The esul s o he e alua ion
using he silhoue e can be seen in Table 5, he g aph is
a ached in Figu e 3, and he clus e dis ibu ion wi h 7
clus e s is shown in Figu e 4.
Table 5. Silhoue e Coe icien e alua ion esul s
Clus e
Silhoue e Sco e
2
0,2627
3
0,4572
4
0,4339
5
0,3917
6
0,5225
7
0,5494
8
0,5349
9
0,5395
10
0,5299
“Clus e ing o Sc ew P ess Machine Condi ions using he K-Medoids Me hod”
7920
Volume 10 Issue 11 No embe 2025ETJ ,
1
M. Tau ik Ap inaldo
Figu e 3: Silhoue e Coe icien e alua ion g aph
Figu e 4: Clus e dis ibu ion wi h k=7
The Silhoue e Coe icien app oach was used o assess
he quali y o clus e ing wi h di e en numbe s o clus e s (K
= 2 o K = 10). The es esul s showed ha he highes
Silhoue e Coe icien alue was ob ained a K = 7 wi h a
sco e o 0.5494, while he lowes alue was ound a K= 2
wi h a sco e o 0.2627. Based on he e alua ion c i e ia, a
Silhoue e Coe icien alue ange be ween 0.5 and 0.7
indica es ha he clus e ing s uc u e o med is in he
ai /good ca ego y, e en hough i is no comple ely
sepa a ed.
b. Da ies-Bouldin Index
The nex e alua ion s age was conduc ed using he
Da ies-Bouldin Index (DBI) me hod. This index p o ides a
quan i a i e measu e o clus e quali y by calcula ing he
a e age le el o simila i y be ween clus e s. The DBI alue is
ob ained om a compa ison be ween he dis ance be ween
clus e cen e s and he dis ibu ion o da a wi hin he clus e .
The smalle he DBI alue, he be e he clus e ing quali y,
as i indica es ha he da a wi hin a clus e is su icien ly
dense and clea ly sepa a ed om o he clus e s. The esul s o
es ing using DBI can be seen in Table 6 and he g aph o he
DBI me hod is a ached in Figu e 5.
Table 6. DBI e alua ion esul s
Clus e
DBI Value
2
1,0537
3
0,7719
4
0,722
5
0,8586
6
0,6157
7
0,5521
8
0,5529
9
0,5635
10
0,5707
Figu e 5: Da ies-Bouldin Index e alua ion g aph
Based on he DBI calcula ion esul s o he numbe o
clus e s (K) be ween 2 and 10, he index alue a ia ions a e
shown in Table 7 and Figu e 4 abo e. The highes DBI alue
occu s a K = 2 wi h 1.0537, while he lowes DBI alue is
ob ained a K = 7 wi h 0.5521. This indica es ha he
con igu a ion wi h se en clus e s p oduces he mos op imal
sepa a ion, because each clus e is su icien ly compac
wi hin i sel and has a ela i ely la ge dis ance om o he
clus e s. Thus, he bes numbe o clus e s acco ding o he
Da ies-Bouldin Index e alua ion is K = 7.
F. Bes Clus e Resul
The op imal numbe o clus e s was ound a k=7 based on
he e alua ion indings u ilizing he Silhoue e and Da ies
Bouldin Index (DBI) me hodologies. The highes silhoue e
alue and lowes DBI bo h consis en ly indica e op imal
clus e sepa a ion quali y. The e o e, u he analysis uses 7
clus e s as he bes con igu a ion.
Table 7. Range o alues o each clus e
Clus e
Min
Tempe a u
e
Max
Tempe a u
e
Min
P essu
e
Max
P essu
e
0
121
155,50
2,10
3,36
1
92,5
145,25
1,15
2,58
2
0
0
0,02
2,32
3
0
0
2,36
3,32
4
100
157
3,16
4,43
5
98
114
2,59
3,33
6
100
133,25
0
1,21
Table 7 shows he empe a u e and p essu e anges o each
clus e . F om he able, i can be seen ha Clus e 0 has a high
“Clus e ing o Sc ew P ess Machine Condi ions using he K-Medoids Me hod”
7921
Volume 10 Issue 11 No embe 2025ETJ ,
1
M. Tau ik Ap inaldo
empe a u e ange o 121–155.5°C wi h mode a e p essu e o
2.10–3.36. Clus e s 1 and 4 ha e ela i ely high empe a u es
eaching 155.5–157°C wi h conside able p essu e, whe e
Clus e 4 shows maximum ope a ing condi ions wi h p essu e
up o 4.43. Meanwhile, Clus e s 2 and 3 ha e a empe a u e
alue o 0 and e y low p essu e, e en eaching 0, wi h
p essu e a ia ions o 0.02-2.32 and 2.36-3.32, which
indica es ha he p essu e in Clus e 3 is highe . O he
clus e s show a ia ions in ange ha desc ibe di e en
ope a ing condi ions. Fo example, Clus e 5 shows s able
ope a ion a a medium le el wi h empe a u es o 98–114°C
and p essu es o 2.59–3.33, while Clus e 6 has no mal
empe a u es o 100–133.25°C bu low p essu e.
CONCLUSIONS
This s udy success ully g ouped sc ew p ess machine
condi ions using he K-Medoids me hod wi h a da ase o
23,002 eco ds om PT. XYZ o he pe iod Ap il–May
2024. The esea ch p ocess began wi h da a selec ion, p e-
p ocessing, ans o ma ion using Z-Sco e No maliza ion,
implemen a ion o K-Medoids clus e ing, and e alua ion
using he Silhoue e Coe icien and Da ies-Bouldin Index
(DBI). The e alua ion esul s showed ha he bes numbe o
clus e s was ob ained a K=7, wi h he highes Silhoue e
Coe icien alue o 0.5494, which is included in he
ai /good clus e ing s uc u e ca ego y, and he lowes DBI
alue o 0.5521, which indica es op imal clus e sepa a ion
quali y. In o he wo ds, he se en-clus e con igu a ion
p oduced he mos ep esen a i e and s able da a sepa a ion
compa ed o o he clus e numbe a ia ions. Fu he analysis
o he empe a u e and p essu e anges in each clus e shows
ha each clus e desc ibes di e en machine ope a ing
condi ions, anging om idle condi ions wi h almos ze o
p essu e, no mal ope a ion in he medium empe a u e ange,
o maximum ope a ing condi ions wi h high empe a u es and
p essu es. These indings indica e ha he K-Medoids me hod
can be used e ec i ely o moni o he condi ion o sc ew
p ess machines.
REFERENCES
1. В. М. Корендій and В. Б. Гавран, “Analysis o he
oil ex ac ion p ocess and p ospec s o au oma ion o
sc ew p ess ope a ion,” Sci. Bull. UNFU, ol. 34, no.
1, pp. 85–90, 2024, doi: 10.36930/40340112.
2. F. Pohan, I. Sapu a, and R. Tua, “Scheduling
P e en i e Main enance o De e mine Main enance
Ac ions on Sc ew P ess Machine,” J. Ris. Ilmu Tek.,
ol. 1, no. 1, pp. 1–12, 2023,
doi: 10.59976/ju i . 1i1.4.
3. S. Yasi and A. Sapu a, “Analisa Reliabili y Dan
A ailabili y Mesin Sc ew P ess Kelapa Sawi (S udi
Kasus di PT. Ujong Neubok Dalam),” Unis ek, ol.
9, no. 2, pp. 83–94, 2022, [Online]. A ailable:
h p://ejou nal.unis.ac.id/index.php/UNISTEK
4. I. Anwa , A. A dal, D. Denu , and D. De mawan,
“Analisis Penyebab Ke usakan Hub Sc ew P ess dan
Op imasi Teknik Pemeliha aan Te hadap Mesin
Pab ik Pengolahan Kelapa Sawi ,” J. Su ya Tek.,
ol. 10, no. 1, pp. 733–737, 2023,
doi: 10.37859/js . 10i1.5003.
5. C. To o a and F. Palumbo, “Clus e ing mixed- ype
da a using a p obabilis ic dis ance
algo i hm[Fo mula p esen ed],” Appl. So Compu .,
ol. 130, p. 109704, 2022,
doi: 10.1016/j.asoc.2022.109704.
6. S. Elka eb, A. Mé walli, A. Shendy, and A. E. B.
Abu-Elanien, “Machine lea ning and IoT – Based
p edic i e main enance app oach o indus ial
applica ions,” Alexand ia Eng. J., ol. 88, no.
Janua y, pp. 298–309, 2024,
doi: 10.1016/j.aej.2023.12.065.
7. M. R. Ka im e al., “Deep lea ning-based clus e ing
app oaches o bioin o ma ics,” B ie . Bioin o m.,
ol. 22, no. 1, pp. 393–415, Jan. 2021,
doi: 10.1093/bib/bbz170.
8. J. Jas il, M. F. Al Fiq i, S. Sanjaya, L. Handayani,
and F. Insani, “Pengelompokan Da a Kondisi Mesin
Sc ew P ess Menggunakan Algo i ma Fuzzy C-
Means,” In . Sys . J., ol. 8, no. 01, pp. 60–70, 2025,
doi: 10.24076/in osjou nal.2025 8i01.2133.
9. F. K. Rahman, J. S. Sanjaya, L. Handayani, and F.
Insani, “Pene apan Algo i ma K-Means Clus e ing
pada Kine ja Mesin Sc ew p ess,” Bull. In .
Technol., ol. 6, no. 2, pp. 59–70, 2025,
doi: 10.47065/bi . 5i2.1783.
10. N. Su eja, B. Chawda, and A. Vasan , “An imp o ed
K-medoids clus e ing app oach based on he c ow
sea ch algo i hm,” J. Compu . Ma h. Da a Sci., ol.
3, no. July 2021, p. 100034, 2022,
doi: 10.1016/j.jcmds.2022.100034.
11. A. Sob inho Campolina Ma ins, L. Ramos de
A aujo, and D. Rosana Ribei o Penido, “K-Medoids
clus e ing applica ions o high-dimensionali y
mul iphase p obabilis ic powe low,” In . J. Elec .
Powe Ene gy Sys ., ol. 157, no. Feb ua y, 2024,
doi: 10.1016/j.ijepes.2024.109861.
12. Al ia Nu laili Tahiya , Bima Maulana, Ade Eka
Sapu a, Lusiana E izoni, and Rahmaddeni
Rahmaddeni, “Klas e isasi Lagu Popule dan
Eksplo asi Subgen e Spo i y 2024 dengan K-
Medoids,” J. In o m. Dan Tekonologi Kompu ., ol.
5, no. 1, pp. 34–48, 2025,
doi: 10.55606/ji ek. 5i1.5699.
13. J. Inayah, A. Fanani, and W. D. U ami, “Klas e isasi
Da a Kejadian Gempa Bumi di Indonesia
Menggunakan Me ode K-Medoids,” J. Sis . dan
Teknol. In ., ol. 12, no. 2, p. 271, 2024,
doi: 10.26418/jus in. 12i2.73594.
14. M. Mina ni, M. Mahend a, A. Anisya, D. W. T.
Pu a, G. Y. Swa a, and I. Wa man, “Klas e isasi
“Clus e ing o Sc ew P ess Machine Condi ions using he K-Medoids Me hod”
7922
Volume 10 Issue 11 No embe 2025ETJ ,
1
M. Tau ik Ap inaldo
Wilayah Rawan Keke asan Anak Menggunakan
Algo i ma k-Medoids di Ko a Padang,” J. Min o
Polgan, ol. 13, no. 2, pp. 2309–2319, 2025,
doi: 10.33395/jmp. 13i2.14447.
15. D. Si anggang, M. Kom, and A. Ap io i, Buku
Monog a Algo i ma Ap io i. 2023.
16. S. E. Saqila, I. P. Fe ina, and A. Iskanda , “Analisis
Pe bandingan Kine ja Clus e ing Da a Mining
Un uk No malisasi Da ase ,” J. Sis . Kompu . dan
In o m., ol. 5, no. 2, p. 356, 2023,
doi: 10.30865/json. 5i2.6919.
17. E. He man, K. E. Zsido, and V. Feny es, “Clus e
Analysis wi h K-Mean e sus K-Medoid in
Financial Pe o mance E alua ion,” Appl. Sci., ol.
12, no. 16, 2022, doi: 10.3390/app12167985.
18. U. Lina i, A. Rahmawa i, A. Hend i Soleliza Jones,
L. Zah o un, and U. Ahmad Dahlan, “Pene apan
Me ode K-Medoids Guna Pengelompokan Da a
Usaha Mik o, Kecil dan Menengah (UMKM)
Bidang Kuline Di Ko a Yogyaka a,” J. Ilmu
Kompu . dan Sis . In ., ol. 7, no. 1, pp. 37–45, 2024.
19. K. Dbscan and Y. Hasan, “Penguku an Silhoue e
Sco e dan Da ies-Bouldin Index pada Hasil Clus e
K-Means dan DBSCAN,” ol. 06, no. 01, pp. 60–74,
2024.
20. H. Hende i e al., “Op imiza ion o Da ies-Bouldin
Index wi h k-medoids algo i hm,” AIP Con . P oc.,
ol. 3065, no. 1, 2024, doi: 10.1063/5.0225220.
21. M. D. Ka ikasa i, “Sel -O ganizing Map
Menggunakan Da ies-Bouldin Index dalam
Pengelompokan Wilayah Indonesia Be dasa kan
Konsumsi Pangan,” Jambu a J. Ma h., ol. 3, no. 2,
pp. 187–196, 2021, doi: 10.34312/jjom. 3i2.10942.
