I ndo ne s i a n  J o u rn a l  o f  E l ect ri ca l  E n g i n eeri n g  a n d  C o m p u t er S ci en ce   V o l.   11 ,  N o.   2 A ug us t   201 8 , p p 48 7 ~ 49 5   I S S N :  2502 - 4752,  D O I :  10. 11 591/ i j e e c s . v 1 1 .i 2 .p p 487 - 4 95          487       Jou r n al  h om e p age h ttp : //ia e s c or e . c om / j our nal s / i nde x . php/ i j e e c s   A T iered  Appro a ch o n Di m e ns io na l Reduc t io n P ro ce s s  f o P redic t io n o f   Co ro na ry  H ea rt  D is e a s e       W i h ar t o W i h ar t o 1 ,  H eri a n t o  H eri a n t o 2 ,  H a r i  K us na nt o 3   1 D ep ar t m en t  o f  I n f o r m at i c,  S eb el as  M ar et  U n i v er s i t y ,  I n d o n es i a   2 D ep ar t m en t  o f  M ech an i cal  &  I n d u s t r i al  E n g i n ee r i ng ,  G a dj a h M a d a  U ni v e r s i t y ,  I ndo ne s i a   3 D ep ar t m en t  o f  M ed i ci n e,  G ad j ah  M ad a U n i v er s i t y ,  I n d o n es i a       A rt i cl e I n f o     AB S T RAC T     A r tic le  h is to r y :   R ecei v ed   No v   24 ,  201 7   Re v i se d   J a n 9 ,  201 8   A ccep t ed   Ap r   21 ,  20 1 8       T he  us e  of  di m e ns i ona l  r e duc t i on  i n t he   di a g nos t i c s y s t e m   m o d el  o f  co r o n ar y   h ear t  d i s eas e,  m an y  s a m e o f  c as e d o  n o t  t ak e i n t o  acco u n t   t h e cl i n i cal   pr oc e dur e s  c om m onl y  us e d b y   c l i ni c i a ns  i n d i a g nos i s .  T h is  r e q u ir e s  th a t th e   e x a m i na t i on be   do ne  t h or o ug hl y ,  t h us  m a k i ng  t he  hi g h c os t  of   di a g nos i s .   T h i s st u dy  a i m s  t de v e l op a  t i e r e d a ppr oa c h m ode l  i n r e duc i ng  d i m e ns i ons   f or  pr e di c t i ng  C H D .   T he   m e t hod i n t hi s  r e s e a r c h i s  di v i de d i nt o s e v e r a l   s t a g e s ,  na m e l y  pr e pr oc e s s i ng ,   bui l di ng  t he  k n ow l e dg e  ba s e  a nd s y s t e m   t es t i n g .  P r e p r o ces s i n g  co n s i s t s  o f  s ev er al  p r o ces s es ,  n a m el y  t h e r e m o v al  o f   m i s s i ng  v a l ue  da t a ,  g r oupi ng  a t t r i but e s ,  a n d di v i d i ng  da t a  f or  t r a i ni ng  a nd   te s tin g .  K n o w le d g e  b a s e   m o d e lin g  is  d iv id e d  in to   th r e e  le v e ls .  T h e  f ir s t le v e l   w er e t h e r i s k   f act o r  at t r i b u t es ,  t h e  s eco n d  l ev el  w er e t h e t y p e o f  ch es t  p ai n  &   E C G ,  a nd t he  t hi r d w e r e  s c i nt i g r a phy  &  c or ona r y  a ng i og r a phy .   T he   k now l e dg e  ba s e   w a s   m ode l e d ba s e d o n f uz z y  r ul e s  a nd i t s  i nf e r e nc i ng   pr oc e s s  us i ng  M a m da ni  m e t hod.   T he  f i r s t ,   f uz z y  r ul e - ba s e w a s   obt a i ne d by   us i ng  t he  F R S  s t udy .  T he  s e c ond a nd t hi r d s t a g e ,  us i ng  t he  i nd uc t i o n r ul e   a lg o r ith m  to  g e t th e  r u le ,   th e n  c o n v e r te d  to  f u z z y  r u le .  T h e  te s te d  a lg o r ith m   w e r e  C 4. 5,  C A R T ,   a nd F D T .   T he  s y s t e m  t e s t i ng   w a s  pe r f or m e d by  t he  5 - f ol ds  c r os s - v a l i da t i on m e t hod,  w i t h pe r f or m a nc e  pa r a m e t e r s  ba s e d on   pop ul a t i on   a n d  in d iv id u a l.  T h e  te s t r e s u lte d  u s i n g  th e  C le v e la n d  a n d   H u n g ar i an  d at as et s ,  t h F R S +C A R T  co m b i n at i o n  w as  cap ab l e o f  r ed u ci n g   t h e m o s t  at t r i b u t es  an d  t h h i g h es t  l i k el i h o o d  r at i o  p er f o r m an ce  p ar am et er ,   w h i ch  w as  1 5 . 9 6 .  F R S +C 4 . 5 ,  at   l eas t  t h e at t r i b u t es  w e r e  r e duc e d,  but   ha s  a A U C  pe r f or m a nc e  o f  80. 4 3% ,  w hi l e  F R S + F D T ,  m or e  r e duc e d a t t r i but e s  t ha F R S +C 4 . 5 ,  an d  A U C  p er f o r m an ce p ar a m et er s  ar e b et t er  t h an  F R S +C A R T .   D i m e ns i ona l  r e duc t i on m ode l  f or  pr e di c t i o n of  C H D ,  c a pa bl e  o f  pr ov i di ng   b et t er  p er f o r m an ce t h an   n o t  t i er ed .   Ke y wo rd s :   D i m e n s io n a l R e d u c tio n   D ia g n o s is   C o r o n ar y  h ear t  d i s ea s e   L i k e li h o o d  r a tio   C opy r i g ht   ©  201 8   I ns t i t ut e  o f  A d v anc e d E ngi ne e r i ng  an Sc i e nc e   A l l  ri g h t s re se rv e d .   Co rre sp o n d i n g  Au t h o r :   W ih a r to  W ih a r to ,   D ep ar t m en t  o f  I n f o r m at i c,   Se b el as  M ar et  U n i v er s i t y ,  I n d o n es i a .   E m a il:  w i h ar t o @ s t a f f . u n s . ac. i d       1.   I NT RO D UCT I O N   T h e s t at e m en t   f r o m  W H O  t h at  car d i o v as cu l ar  d i s eas e o f  t h e h ear t  an d  b l o o d  v es s el s ,  es p eci al l y   co r o n ar y   h ear t   d i s eas e   ( C H D )   w a s   s t i l l   r a n k ed   a s   t h l ead i n g   cau s o f   d eat h   i n   d e v e lo p in g   c o u n tr ie s   u n til   2020.   T h e  di s e a s e  c a n  be  pr e v e n t e d  b y   m a k i n g a n  e a r l y   di a g n os i s .  U nf or t un a t e l y ,  t h e  c os t  o f  a  t h or oug ex a m i n at i o n  f o r  t h e d i ag n o s i s  o f  h ear t  d i s eas e i s  r el at i v e l y   ex p en s i v f o r  t h e s i ze o f  t h d ev el o p i n g  co u n t r y .   T he  hi gh c o s t  o f   s uc h d i a g no s i s  b r i ng s  a n e c o no m i c  i m p a c t   o n a  c o unt r y ,  na m e l y t he  d e c l i ne  i n p r o d uc t i v i t o f  th e  p o p u la tio n ,  a  c o u n tr y .  T h e s e  c o n d itio n s  e n c o u r a g e  t h e  d e v e lo p m e n t o f  c lin ic a l d e c is io n  s u p p o r t s y s te m   m o d el s  f o r  t h e d i ag n o s i s  o f   ar t i f i ci al  i n t el l i g e n ce b as ed   c o r o n ar y  ar t er y  d i s eas e,   w h i c h  t ak es  i n t o  acco u n t   cl i n i cal  co s t s  an d  p r o ced u r es .   Evaluation Warning : The document was created with Spire.PDF for Python.
                      I SSN :   25 02 - 4752   I nd o ne s i a n J  E l e c  E ng  &  C o m p  S c i ,   V o l.   11 , N o .   2 A ug us t  2018   :   4 87     495   488   T h e  s y s t e m   m ode l  of  di a gnos i s  on  c or on a r y  h e a r t  di s e a s e  i s   w i de l y  de v e l ope d by  u s i n g  a   c o m b i na t i o n o f  d i m e ns i o na l  r e d uc t i o n a nd  c l a s s i f i c a t i o n.  D i m e ns i o na l  r e d uc t i o n i n t he  d ia g n o s tic  s y s te m   m ode l  i s  di v i de d i n t o t w o,  n a m e l y  di m e n s i o n a l  r e du c t i on  t a k i ng  a c c oun t  of  c os t  a n d n ot  c on s i de r i ng  c os t .  T h e   m o d el  o f  t h e d i a g n o s t i c s y s t e m   w i t h   n o  co s t  co n s i d er at i o n ,  h as  b een  l ar g el y  d o n e f r o m   t h e t i m e p er i o d  f r o m   2009 - 2 0 1 7 .  R es ear ch  co n d u ct ed  b y   L i u  et . al   [1 ]   pr opos e d t h e  di m e n s i o n a l  r e du c t i on   m o de l ,  by  c o m bi n i ng  R e lie f F  a n d   H e u r is tic   R o u g h   S e ts   a l g o r ith m s .   T h e   s t u d y   h a s   a   n u m b e r   o f   w e a k n e s s e s .   F ir s t,   t h e   a ttr ib u te s   ge ne r a t e d  i n t he  d i m e n s i o na l  r e d uc t i o n p r o c e s s ,  w he n r e f e r r i ng t o  t he  c o s t i n g gr o up i ng s  o f  F e s hk i  a nd   S hi j a ni ' s   [2 ]   r es ear ch   r es u l t s ,   t h er ar t w o   co s t l y   c h ec k s ,   n a m e l y   s ci n t i g r ap h y   an d   f l u o r o s co p y .   A   s i m i l ar   r es u l t  i s  al s o   m ad e b y   A h m ad i  et . al   [3 ] ,  a co m b i n at i o n   m o d el  o f  C 5 . 0   w i t h  a  n eu r a l   ne t w o r k,   w hi c h d o e s  no t   r e du c e  bot h  t h e  c os t l y  a t t r i bu t e s .   S e c on d,  di m e n s i o n a l  r e du c t i on  pr oc e s s  doe s   n ot  t a ke  i n t o a c c oun t  t h e   cl i n i cal  p r o ced u r es  t h at  cl i n i c i an s   n o r m al l y  p er f o r m .   A  s y s te m  o f  d ia g n o s tic s  w it h  c o s t - ba s e d di m e ns i on r e du c t i on  h a s  be e n  pe r fo rm e d  b y     A r j en ak i  et . al   [4 ]   a nd  a l s o  b F e s h ki  a nd  S hi j a ni   [2 ] .  B ot h   of  t h e s e  s t u di e s  i n di m e ns i on a l  r e du c t i on  u s i n g a   g e n e t ic  a l g o r ith m  a n d  p a r tic le   s w a r m  o p ti m iz a tio n   w i th   its  f i tn e s s   f u n c tio n  i s  a  c o s t  f u n c tio n .  B o th   s tu d ie s  a r e   cap ab l e o f  r ed u ci n g  co s t l y  at t r i b u t es .  U n f o r t u n at el y  t h m o d el  d o es  n o t   c o n s id e r  th e  s ta g e s  o f  e x a m in a tio n   a ttr ib u te s ,  c o n s e q u e n tl y  a t t h e   ti m e  o f  th e  d ia g n o s is  p r o c e s s  r e q u ir e s  e x a m i n a tio n  o f  a ll a ttr ib u te s  d i m e n s io n a l   r e d uc t i o n r e s ul t s .  T hi s   m a ke s  t he  a b s e nc e  o f  a  r e d uc t i o n p r o c e s s  a ga i w h e n t he   d i a gno s i s  i s  d o ne .   D i me n s i o n al  r ed u c t i o n   m o d el s  cap ab l e o f  r ed u ci n g  co s t l y  at t r i b u t es   h a v e al s o  b e en   u n d er t a k en  b y     W ih a r to  e t. a [5 ] .  T h e s t u d y   u s ed  a co m b i n at i o n  o f  t i er ed  ap p r o ach es   w it h  lo g i s tic  r e g r e s s io n  a lg o r it h m s ,   u n f o r t u n at el y ,  t h e t i er ed  ap p r o ach  i s  o n l y   u s ed  i n  d i m en s i o n al  r ed u ct i o n  p r o ces s ,  b u t   at  t h e t i m e o f  t h di a g n os i s  pr oc e s s  i s   n ot  u s e d,   s o n o di m e ns i on a l  r e du c t i on  o c c u r s   w he n  t h e  di a gn os i s  s y s t e m  i s   u s e d.   T he  us e  o f  t i e r ed  co n cep t s  h a s  al s o  b een  d o n e  b y  W i h ar t o  et . al   [6 ] .   T he  s t ud y   gr o up e d  t he  a t t r i b ut e s   a c c o r d i ng t o  t he  s t a ge s  i n t he   p r o c e s s  o f  d i a gno s i s ,  b ut  i n e a c h s t a ge   w i t ho ut   di m e n s i o n a l  r e du c t i on  pr oc e s s .   D i m e ns i o na l   r e d uc t i o p r o c e s s   o c c ur s   o nl w he t he   d i a gn o s t i c   p r o c e s s   i s   d o ne ,   i e   i f   t h e   d i a gno s i s   i t h e   ear l y  s t a g es  h a s  b een  d ecl ar ed  n eg at i v e,  n o  f u r t h er  ex a m i n at i o n  o f  t h e at t r i b u t es  at  t h e n e x t  s t ag e i s  r eq u i r ed .   T he  s am e co n cep t  h a s  al s o  b een  u s ed  b y  W i h ar t o  et . al   [7 ] ,  th e  s tu d y   u s e d  th e  C 4 . 5  a lg o r ith m   f o r  in d u c tio n  o f   th e  r u le  a s   w e ll a s  d i m e n s io n a l r e d u c tio n .  U n f o r tu na t e l y ,  i n t he  d i m e n s i o n r e d uc t i o n r e s ul t s  ge ne r a t e d   w he n   u s i n g  t h C 4 . 5 ,  o n l y  o n e at t r i b u t e i s  r ed u ced .  T h i s   m ak e s  t h m o d el  l e s s  ab l e t o  r ed u ce t h e co s t  o f  t h e   e xa m i na t i o n,  e s p e c i a l l y a t t r i b ut e s  t ha t  c o s t l y i n t he  e xa m i na t i o n.   A no t he r  d i s a d v a nt a ge   i s  t ha t  t he   pe r f or m a n c e   m e a s u r e m e n t  i s  on l y  popu l a t i on - b as ed ,   w h e r eas  i n d i v i d u al - b as ed   [8 ]   w it h  li k e li h o o d  r a tio   p ar am et er s  h as   n o t  b een  d o n e,  b u t  t h e t es t i n g   m et h o d  d o es  n o t  u s e k - f ol ds  c r os s - v a l id a tio n .   R e f e r r i n g  t o a  num be r  of   s t u d i es  t h at   h av e b een  d e v el o p ed ,   m o s t  d i m e n s i o n al  r ed u ct i o n  o ccu r s  o n l y   w h e n  p r io r  to  c la s s i f ic a tio n ,   w h ile  t h e  d i m e n s io n a l r e d u c t io n  a t t h e  ti m e  o f  t h e  d ia g n o s tic   s y s te m   w h ic h   u s e d   w a s  s till  m i n i m a l.  I n  a d d itio n ,  th e s e  s t u d ie s  o f  r e d u c e d  a ttr ib u te s   w e r e   s till s m a ll,   r e s u lti n g  in  c o s tl y   co n s eq u e n ces  t h at   m u s t  b e i n cu r r ed  i n  t h e d i ag n o s t i c p r o ces s .  F u r t h er m o r e,  p r ev i o u s  s t u d i es   h av m o s t l y   c on du c t e m e a s u r e m e n t s  o f  popu l a t i on - b as ed  o n  p er f o r m a n ce.  T h er e w a s  s t i l l   m i n i m al  i n d i v i d u al - ba s e d on  p er f o r m a n ce  m ea s u r e m e n t,   th a lik e l ih o o d   r a tio   p a r a m e t e r .  B a s e d   o n   th is ,   in   th i s   s tu d y   p r o p o s e d   a   di m e ns i on a l  r e du c t i on  m ode l   t h a t  oc c u r r e w h e n  di m e ns i ona l  r e du c t i on pr oc e s s  a n d di a gn os t i c  pr oc e s s .  T h i s   m ode l  c a n   m a k e  t h e  s y s t e m   m ode l  m or e  pr ov i de  c os t  s a v i n gs  i n  t h e  di a g n o si s  o f  CH D .         2.   R ES EA R C H  M ETH O D   2.     M at e r i al  an d  D at a   T h e  s t u d y   w a s  de v e l ope d u s i ng  t h e   C l e v e l a n d a n d H ung a r i a n  da t a s e t s  of  t h e  U C I  r e pos i t or y ,   w hi c can  b e acces s ed  o n l i n e   [9 ] .   T h e  da t a s e t  u s e d c ons i s t s  of  14 a t t r i bu t e s ,   w hi c h  c o n s i s t  of  13 i n de pe n de n t   a ttr ib u te s  a n d  1  d e p e n d e n t  a ttr ib u t e.  T h e co m p l et e at t r i b u t es  ar s h o w n  i n  T ab l e 1 .  T h e n u m b er  o f  t h e   C l ev e l an d  d at as et s   w a s  3 0 3 ,   w h i l e t h e H u n g ar i a n  2 9 4 .  T h e d at as et  h as  t w o  o u t p u t s ,   h eal t h y   w h i ch   s y m bol i z e d by  0,  a n d s i c k  s ym bol i z e d b y  1.                         Evaluation Warning : The document was created with Spire.PDF for Python.
I nd o ne s i a n J  E l e c  E ng  &  C o m p  S c i     I SSN :   2502 - 4752       A  T i e r e d A ppr oac on D i m e ns i onal  R e d uc t i on P r oc e s s  f or  P r e di c t i on of  ( W i har t o W i har t o )   489   T ab l 1 .  A ttr ib u t es  i n  t h e d i ag n o s i s  o f  co r o n ar y  h ear t  d i s ea s e   No   A ttr ib u te s   D e s c r ip t io n   1   a g e   Ag e   2   s ex   G en d er   3   cp   C h e s t  p a i n   t y p e   4   r e s t b p s   R e st i n g  sy st o l i c  b l o o d  p r e ssu r e   5   c h ol   C h o l e s t e r ol  i n  m g / d l   6   fb s   F a s t i n g  b l o o d   s u g a r   7   r es t ec g   Re s t i n g  E CG   8   t h al ac   M ax i m u m  h ea r t  r a t e ach i ev ed   9   ex a n g   E xe r c i s e   i n d u c e d  a n gi n a   1 0   o l d p e a k   S T  d ep r es s i o n  i n d u c ed   b y  ex er c i s e r e l a t i v e t o  r es t     1 1   s lo p e   T h e s l o p e o f  t h e S T  s eg m en t  f o r  p ea k   ex er c i s e   1 2   ca   N u m b e r  of  m a j o r  ve s s e l s  c o l or e d  b y  a n g i o gr a p h y  C o r o n a r y  ( 0 - 3 )   1 3   t h a l   D ef ec t  t y p e  ( S c i n t i g r a p h y )   1 4   o u t p ut   L ev el  H e ar t  d i s e as e  ( 0 / 1 )       2.       M e t ho d   T h e   m e t h od u s e i n t hi s   s t u dy   w a s  a s  s h o w n i n  F i gu r e  1,   w hi c h   w a s  di vi de d i n t o t h r e e   pa r t s .  T h e   f i r s t  i s  p r ep r o ces s i n g  t h at  p er f o r m s  t h m i s s i n g  v al u e d at a r e m o v al  p r o ces s ,   a ttr ib u te s  g r o u p in g ,  d a ta  s h a r in g   in to  tr a i n in g  a n d  te s ti n g  d a ta ,   w i th  5 - f ol ds  c r os s - v a l i da t i on  m e t h od.  T h e  g r ou pi ng  of  a t t r i bu t e s   w a s  di vi de i nt o   t hr e e   gr o up s .   T he   f i r s t   g r o up s   a r e  r i s f a c t o r s ,   b o t c he s t  p a i &   E C G ,   a nd   s c i nt i gr a p h y   &  c o r o na r a ngi o gr a p h y .  T h s eco n d   i s   t h p r o ces s  o f   b u i l d i n g   k n o w l ed g b as e  f o r  eac h   a t t r i b u t e g r o u p .   T h t h r ee   g r o u p s  at  t h e s a m e t i m e r ef l ect  t h e e x a m i n at i o n  l e v el .  D ev el o p m e n t  o f   k n o w l ed g e b as e f o r  l ev el  1 ,  b y   m ode l i ng  t a bl e  s c or e  i n   s t u d y  F r a m i ng h a m  r i s k s c or e ( F R S ) [ 10] ,  i nt o  f uz z y r ul e - b a s e d .  I n t he  s e c o nd  a nd   t h i r d l e v e l ,  i t   w a s  pr e c e de d b y  i n d u c t i on pr oc e s s   u s i ng   n on - b l ack - box c l a s s i f i c a t i on a l g or i t hm .  T h e  a l g or i t hm   we r e   C 4 . 5 ,  C A R T ,  a nd  F D T .  A f t e r  i nd uc t i o n o f  t he  r ul e  t h e n c o n ve r t e d  t o   f uz z y r ul e - ba s e d.  T h i r d,  pe r f or m   t he   t e s t i n p r o c e s s   w i t t he   m o d e l ,   a s   s ho w i F i g ur e   1 .   A t   e a c l e ve l   i m a ki ng  a   d e c i s i o u s i ng  f uz z i n f e r e nc e  s ys t e m   w i t h M a m d a ni   m e t ho d .   T h e p er f o r m a n c e  of  t h e  pr opo s e d di a g n o s t i c  s y s t e m  us e s   t w o a ppr oa c h e s ,  n a m e l y  popu l a t i on - b as ed   a nd  i nd i vi d ua l - ba s e d.  T h e  popu l a t i on - ba s e d pe r f or m a n c e ,   u s i ng  a  num be r  of  pa r a m e t e r s ,  i s  r e f e r r i ng  t o t h e   m at r i x  co n f u s i o n  t ab l e,  as  s h o w n  i n  T ab l e 2 .  T h e p ar am et er s  ar e s e n s i t i v i t y ,   s p eci f i c i t y ,  accu r ac y ,  a n d  ar ea   und e r  t he  c ur ve  ( A U C )   [ 11] .       T ab l 2 .  C o n f u s io n  M a tr ic   A ct u al  C l as s   P r e d i c ti o n  C la s s   P o s itiv e   N e g a t i ve   P o s itiv e   T P  (T ru e  P o s i t i f)   F N  ( F al s e  N eg at i v e )   N e g a t i ve   F P  ( F a l s e  P o s itif )   T N  ( T r u e N eg a t i v e)       T h e s eco n d  ap p r o ach   w as   i n d i v i d u al - b as ed ,  i e p er f o r m an ce p ar a m et er s  t o  i n t er p r et  p at i en t s   in d iv id u a ll y ,  ie  b y   th e   lik e lih o o d  r a tio  p a r a m e te r   [8 ] .  R e f e r r i ng  t o  t he  l i ke l i ho o d  r a t i o ,  i t  c a n t he n b e   us e d  t o   cal cu l at e t h e p r e - t e s t  a n d pos t - t e s t  p r o b ab i l i t y .  T h e p er f o r m a n ce p ar a m et er s  can  b e  f o r m u l at ed  i n  t h E q ua t i o n ( 1 - 8 ).     s e n s it iv ity = TP TP + FN                   (1 )     s p e c if ic it y = TN TN + FP                   (2 )     L R + = s e n s itiv ity : ( 1 s p e c if ic ity )             (3 )     L R = ( 1 s e n s itiv ity ) : s p e c if ic ity             (4 )     p re t es t   p r ob a b il ty = TP + FN TP + FN + FP + TN             (5 )     p re t es t   o d d s = p re t es t   p r ob a b il ity : ( 1 p re t es t   p r ob a b il ity )     (6 )     po s t t es t   o d d s = p re t es t   o d d s L R +                          (7 )   Evaluation Warning : The document was created with Spire.PDF for Python.
                      I SSN :   25 02 - 4752   I nd o ne s i a n J  E l e c  E ng  &  C o m p  S c i ,   V o l.   11 , N o .   2 A ug us t  2018   :   4 87     495   490   po s t t es t   p r o ba bi l i t y = po s t t es t   o d d s : ( 1 + po s t t es t   o d d s )                (8 )     2.     I nduc t i o n A l g o r i t h m   A l g or i t hm   C 4. 5 i s  on e  of  t h e  a l g or i t hm s  i n  de c i s i on  t r e e  l e a r n i ng .  T h e  C 4. 5 a l g or i t hm  i s   a   de v e l op m e n t  of  t h e  I D 3 a l g or i t hm   [ 12] .  T h e  I D 3  a lg o r ith m  a tte m p ts  to  b u ild  a  to p - d o w n  d eci s i o n  t r ee,  b y   s p e c if y in g  t h e  r o o t f ir s t.  T h e  r o o t d e te r m i n a tio n  p r o c e s s  is  d o n e  b y  e v a l u a ti n g  a ll a ttr ib u t e s  b y   us i n g a   s t at i s t i cal   m eas u r e ( i n f o r m at i o n  g a i n )  t o   m eas u r e t h e  ef f ect i v en es s  o f  at t r i b u t es  i n  cl a s s i f y i n g   t h s a m p l e d at a.   T h e d i f f er en ce i n  a C 4 . 5  al g o r i t h m   w i t h  I D 3   w as  i n  t h e d e f i n i n g  at t r i b u t es  as  r o o t .  T h e cal cu l at i o n  o f  t h g ai n   i n t he   C 4 . 5  a l go r i t h m  a s  s ho w n i n t he  e q ua t i o ns  ( 9 )  a nd  ( 1 0 )   [ 12]   [ 1 3] .     G a in R a tio ( S , A ) = G ai n ( S , A ) S p l i t I n f o ( S , A )                 (9 )     S p l itI n f o ( S , A ) = S v S l og 2 S v S c i = 1                 ( 10)     T h e  F D T  a l g or i t hm   i s  a n i m pr ov e m e n t  of  t h e   C 4. 5 a l g or i t hm  de v e l ope d by  J i a ng   S a n d H a r r y   Z ha n g   [ 14] .  F D T  h as  a p r o ces s  o f  co n s t r u ct i n g  a   d e c is io n  tr e e  th a t  is  b e tte r  a n d  f a s te r ,   w h ic h  i n d ic a te d  b y  it s   c o m p l e xi t y.   T he   d e c i s i o t r e e   i F D T   i s   b ui l t   us i n i nd e p e nd e nt   i n f o r m a t i o ga i ( I I G ) ,   a s   a   c r i t e r i o f o r   s ep ar at i o n   [ 14] .  I I G   w a s  c a lc u la te d   f o r  a ll c a n d id a te  a ttr ib u te s .  T h e  a ttr ib u te   w it h  t h e   h i g h e s t I I G   v a l u e   w i ll   b e s el ect ed  as  t h e r o o t  n o d e.       Data  colle ction The process of grouping Attribute based on the  inspe ction stage Data proce ssing  Missing Value Data  Testing Data  Training 5 - Fold Cross - Validation Data collection  ( Scintigraphy  Fluoroscopy ) N on - black - box  classification  approach algorithm Rule  Collection Fuzzy Rule   Base d K nowle dge Base Scintigraphy  fluoroscopy Collection of  e xamination  results Chest  pain  ECG N on - black - box  classification  approach algorithm Rule  Collection Fuzzy Rule   Base d K nowle dge Base Chest pain  ECG Collection  of risk  factor data Framingham Risk  Score Study FRS  K nowle dge  Base Fuzzyfication continuous  data  ? Fuzzy Infe rence  Engine Yes Defuzzification Thre shold ? Performance Fuzzy Infe rence  Engine Yes Defuzzyfication Fuzzy Infe rence  Engine Sure  ? Defuzzyfication N o Performance  e valuation Performance  e valuation Performance  e valuation Performance Performance K nowle dge Base Modeling K nowle dge Base Modeling Fuzzy Rule   Base d K nowle dge Base Modeling T he  f ir s P     lev el T he  s ec ond  lev el T he  P hir lev el     F i gu r e  1.  T h e  pr opos e of   m e t h od   Evaluation Warning : The document was created with Spire.PDF for Python.
I nd o ne s i a n J  E l e c  E ng  &  C o m p  S c i     I SSN :   2502 - 4752       A  T i e r e d A ppr oac on D i m e ns i onal  R e d uc t i on P r oc e s s  f or  P r e di c t i on of  ( W i har t o W i har t o )   491   T he  C A R T  a l go r i t h m   w a s  a cl as s i f i cat i o n  al g o r i t h m  b as ed   o n  t h e d eci s i o n  t r ee,  i n t r o d u c ed  b y  L eo   B r e im a n  e t a l.  in  1 9 8 4   [ 15 ] .  C A R T  p r o d u ces  a cl as s i f i cat i o n  t r ee i f  t h e r es p o n s e v ar i ab l w a s  cat eg o r i cal ,  an d   p r o d u ces  a r eg r es s i o n  t r ee i f  t h e r es p o n s e v ar i ab l w as  co n t i n u o u s .  T h e C A R T  al g o r i t h m  h as  t h e   f o llo w in g   st e p s:   a.   P r ep ar e t h e b r an ch  can d i d at es   f o r  al l  p r ed i ct o r  v ar i ab l es ,  eac h  d i v i d ed  i n t o  2 ,  t h e  l ef t  an d   t h e r i g h t  b r a n ch   can d i d at es .   b.   A s s es s  t h e  o v er al l  p r o s p ect i v e b r an c h  t h at  i s  o n   t h e  l i s t  o f  ca n d i d at es   f o r  t h e l a t es t  b r an ch e s .  T h p er f o r m a n ce o f   eac h  b r an ch  c an d i d at w i l l  b m eas u r ed  b y  a  m eas u r e cal l ed  co n f o r m i t y .   c.   D et er m i n w h i ch  b r an c h  can d i d at w i l l  act u al l y  b e a b r an ch .   d.   I f  n o   m o r e  d e c is io n  n o d e s ,  th e  i m p le m e n ta t io n  o f  t h e   C A R T  a lg o r ith m   w ill b e  te r m in a te d .  H o w e v e r ,  i f   th e r e  is  s t ill a  d e c is io n   n ode ,  g o ba c k  t o t h e  s e c on d s t e p,  b y  f i r s t  t h r o w i ng  t h e  br a n c h  c a ndi da t e   w h o h a s   s u cce s s f u l l y  b eco m e a b r an c h .   T h e m ai n  co n cep t  o f  t h e f u zz y  l o g i c t h eo r y   w as  t o  p l ace an  i n p u t  s p ace i n t o  t h e o u t p u t  s p ace u s i n g   IF - T H E N  r u l es ,  a n d  t h e co n c ep t  can  b e u s ed  t o   s o l ve  t he   unc e r t a i nt y p r o b l e m   [ 16] .  M a p p i ng  w a s  d o ne  b us i n g F uz z y I n f e r e nc e   S ys t e m   ( F I S ) .  F I S  e v a lu a te s  a ll r u le s   s i m u lta n e o u s l y   to  g e n e r a te  c o n c lu s io n s .  F I S  h a s   a  c om pon e n t  c ons i s t i n g  o f   f u z zi f i cat i o n ,   m e m b er s h i p  f u n ct i o n ,  i n f er en ce p r o ces s  an d  d ef u zzi f i cat i o n   [ 17] .       3.   RE S U L T S  AND D I S CU S S I O N   3.     R e su l t s   T h e d i m e n s i o n al  r ed u ct i o n   m o d el  w i t h  a t i er ed  ap p r o ach  i n  t h e cas e o f  p r ed i ct i o n  o f  co r o n ar y   h ear t   d is e a s e   w a s  d iv id e d  in to  th r e e  le v e l s ,  as  s h o w n  i n  F i g u r e 1 .  T h e p r o ces s  o f  t h e d i ag n o s i s   w a s  d o n e b y  t i er ed   m o d e l,  i f  a t t h e  le v e l o f  1  d e c la r e d  p r e d ic tio n  r e s u lt s  h a v e  a   h ig h  r is k ,  t h e n   g o  to  th e  le v e o f  2 ,  b u t if   n o t th e n   t h e d i ag n o s i s   w as  co m p l et e.  I f  t h e d i ag n o s i s  r es u l t s  at  t h s eco n d  le v e l,   it is  p o s s ib le  to   m a k e  a  d ia g n o s tic   d e c is io n   f o r  th e  c lin ic ia n ,   th e r e  is  n o   n e e d   f o r  a  d ia g n o s is   a t th e   th ir d  le v e l,  b u t i f   n o t,  t h e n  th e  th ir d  le v e l   d i a gno s i s  i s  d o ne .   A t  l e ve l  1 ,   gi ve s  t he  o ut p ut  o f  t he  p e r c e n t a ge  o f  c o r o na r y he a r t  d i s e a s e  e ve nt s  i n  t he   ne x t   10  y e a r s .  A  f ol l o w - u p  d i ag n o s i s  t o  t h s eco n d  l e v el   w i l l  b e p er f o r m ed  i f  t h e  g e n er at ed   p r o ces s i o n  p as s e s  a  c e r ta in  th r e s h o ld .         F i g u r e 2 .  T h e p er f o r m an ce o f   t h e d i ag n o s t i c s y s t e m  a t  t h e l e v el s  1 ,  2  an d  3       D e te r m i n a tio n   o f   th r e s h o ld ,   w h ic h   w as   t h p er cen t a g o f   t h e p r ed i ct i o n   o f   co r o n ar y   h e ar t   d i s eas r ef er s  t o  s ev er al  co n s i d er at i o n s  t h at   w er e:   a.   t h e i n f l u e n ce o f  a t h r e s h o l d  v a l u e o n  t h e p er f o r m a n ce at  l ev e l  1 ,  l ev el  2  an d  l ev el  3 ,   b.   A t  t he   f i r s t  s t a ge  i s  t he  s c r e e ni ng  s t a ge ,  t h us  e m p ha s i z i n g t h e  s en s i t i v i t y  p er f o r m an ce p ar a m et er   [8 ] ,   c.   r ef er s  t o  t h e p er cen t ag e p r ed i ct ed  i n ci d en ce o f  co r o n ar y  h e ar t  d i s eas e i n  t h n ex t  1 0   y ear s ,  i f  v al u es   ar <1 0 %  s ai d  t o  h av e l o w  r i s k   [1 8 ] .   A   num be r  o f  t h e s e   co n s i d er at i o n s   u s ed  t o  a n al y ze t h e r es u l t s  o f  t es t i n g  at  l e v el  1 .   T es t i n g  t o   d et er m i n e t h r es h o l d  i s  d o n e b y  u s i n g  t r ai n i n g  d at a.  T h e r es u l t i n g  p er f o r m a n ce as   s h o w n  i n  F i g u r e 2 , r ef er r i n g   t o  a n u m b er  o f   t h e s e co n s i d er at i o n s  a n d  t h e p er f o r m a n ce o f  t h e t es t   r es u l t s ,  t h e p er cen t a g e v al u e o f  co r o n ar y   h ear t  d i s eas e  ev e n t s ,   w h i ch   w a s   u s ed  as   a  t h r e s h o l d   w as   >  7 % .  T h is  v a lu e   in d ic a te s   th a t i f  t h e  p r e d ic te d   o u tp u t a t le v e l 1   w a s  >  7 % ,  th e n  th e  d ia g n o s is   w a s  r e q u ir e d  a t th e  s e c o n d  le v e l.   5 5 6 0 6 5 7 0 7 5 8 0 8 5 9 0 9 5 1 0 0 1 2 3 4 5 6 7 8 9 1 0 1 1 1 2 1 3 1 4 1 5 1 6 1 7 1 8 1 9 2 0 %  ( P ercen t )   %   P red i ct ed  I n ci d en ce O f  C o ro n a ry   H ea rt  D i s ea s e   S e n s i t i vi t y- 1 AUC - 2 AUC - 3 Evaluation Warning : The document was created with Spire.PDF for Python.
                      I SSN :   25 02 - 4752   I nd o ne s i a n J  E l e c  E ng  &  C o m p  S c i ,   V o l.   11 , N o .   2 A ug us t  2018   :   4 87     495   492   T ie r e d  s y s te m   m o d e l te s t w a s   don e  by  us i n g  5 - f ol d  c r os s - v a l i da t i on   m e t h od.  P e r f or m a n c e   p ar am et er s  u s ed   w er e p o p u l at i o n - ba s e d a n d i n di v i du a l  ba s e d.   T h e  r e s u l t s  of  t h e  t e s t s  pe r f or m e w he n   us i n t h e t h r es h o l d  at   l ev el  1  ar e   >   7% ,  t h e  r e s ul t i n g popu l a t i on - b as ed  p er f o r m an ce  as  s h o w n  i n  T a b l e  3 .  T he  ne xt   p er f o r m a n ce p ar a m et er   w a s  i n d i v i d u al - b as ed  p ar a m et er s ,   n a m el y  t h e  l i k e l i h o o d  r at i o .  T h e p ar am et er s  can  b u s ed  t o  cal c u l at e  t h e p r e - t e s t   pr oba bi l i t y ,  pos t - t es t  p r o b ab i l i t y ,  a n d  t h e d i f f er e n ce.  T h es v al u e s   in d ic a te  h o w   m uc h i nf l ue nc e  t he   u s e o f  a p r ed i ct i o n   m o d el  o f  co r o n ar y  h ear t  d i s eas e.  T h e r es u l t i n g  p er f o r m a n ce as  s h o w n   in  T a b le  4 .       T ab l 3 .  S y s t e m  pe r f or m a n c e   f or  popu l a t i on - b as ed   f o r  each  l ev el   A l g o r ith m   L ev el   S e n s itiv ity  ( % )   S p e c i f ic it y  ( % )   A ccu r acy  ( % )   AU C   (% )   FR S   1   9 8 , 58   7 , 6 4   5 0 , 67   5 3 , 11   C 4. 5   2   7 5 , 89   8 2 , 17   7 9 , 19   7 9 , 03   3   6 5 , 96   9 4 , 90   8 1 , 21   8 0 , 43   FR S   1   9 8 , 58   7 , 6 4   5 0 , 67   5 3 , 11   CA RT   2   7 1 , 63   7 8 , 98   7 5 , 50   7 5 , 31   3   6 0 , 99   9 6 , 18   7 9 , 53   7 8 , 59   FR S   1   9 8 , 58   7 , 6 4   5 0 , 67   5 3 , 11   FD T   2   8 1 , 56   7 3 , 89   7 7 , 52   7 7 , 72   3   6 3 , 83   9 4 , 9 0   8 0 , 20   7 9 , 37       T h e s y s t e m  p er f o r m a n ce s h o w n  i n  T ab l e 3  an d   T ab l 4  w er e t h e s y s t e m  p er f o r m an ce d er i v ed  f r o m   t h e u s e o f  a n u m b er  o f  d i m e n s i o n  r ed u ct i o n  at t r i b u t es .  D i m en s i o n  r ed u ct i o n  u s i n g   f eat u r e s el ect i o n   m et h o d   t y pe  e m be dde d.   T h e  m e t h od w a s  a l s o u s e d f or  di m e n s i on  r e du c t i on  a n d i n d u c t i on  r ul e .  T h e  a l g or i t hm   f or  t h e   i nd uc t i o r ul e   us e d ,   i nc l ud i n w i t hi t he   d e c i s i o t r e e   f a m i l y.   T he   a l go r i t h m s   a r e   C 4 . 5 ,   F D T ,   a nd   C A R T .   T h e r es u l t  o f  t h e d i m e n s i o n al   r ed u ct i o n  p r o ces s  as  s h o w n  i n   T ab l 5 .       T ab l 4 .  S y s t e m  p er f o r m an ce  f o r  i n d i v i d u al - b as ed  f o r  each  l ev el   A l g o r ith m   L ev el   LR +   P r o b a b il it y   d i ffe re n c e s  (% )   P re - te s t ( % )   P os t - te s t ( % )   FR S   1   1 , 0 7   4 7 , 32   4 8 , 94   1 , 6 3   C 4. 5   2   4 , 2 6   4 7 , 32   7 9 , 26   3 1 , 94   3   1 2 , 94   4 7 , 32   9 2 , 08   4 4 , 76   FR S   1   1 , 0 7   4 7 , 32   4 8 , 94   1 , 6 3   CA RT   2   3 , 4 1   4 7 , 32   7 5 , 37   2 8 , 06   3   1 5 , 96   4 7 , 32   9 3 , 48   4 6 , 16   FR S   1   1 , 0 7   4 7 , 32   4 8 , 94   1 , 6 3   FD T   2   3 , 1 2   4 7 , 32   7 3 , 72   2 6 , 40   3   1 2 , 53   4 7 , 32   9 1 , 84   4 4 , 52       T ab l 5 .   T h e ab i l i t y  t o  r ed u ce  at t r i b u t es  at  eac h  l ev el   A l g o r ith m   l ev e l   th e   a m o u n t is   r ed u c ed   n u m b e r  o f   a t tr ib u t e s   A ttr ib u te s   FR S   1   1   4   a g e,  s ex ,  r es t b p s , ch o l   C 4. 5   2   1   5   c p ,  t h a l a c ,  e x a n g ,  o l d p e a k ,  s l o p e   FD T   2   3   3   cp ,   o l d p e a k , t h al ac   CA RT   2   5   1   cp   C 4. 5   3   0   2   ca, t h al   FD T   3   0   2   ca, t h a   CA RT   3   0   2   ca, t h al       3.      D i s c u ssi o n s   T h e d i m en s i o n a l  r ed u ct i o n   m o d el  w i t h  a t i er ed  ap p r o ach  i n  t h e p r ed i ct i o n  s y s t e m  o f  co r o n ar y  h ear t   d i s eas e,  u s i n g  t w o  ap p r o ach es .  F i r s t ,  t h e d i m e n s i o n al  r ed u ct i o n  ap p r o ach  t o  t h e s y s t e m  as  a  w h o l e a n d   d yna m i c .  T hi s  a p p r o a c h o c c ur r e d   w he n t he  s ys t e m   w a s   us e d  f o r  d i a gno s i s .  T he  s e c o nd ,  d i m e n s i o n r e d uc t i o ap p r o ach  i n  eac h  l e v el ,  e s p eci al l y  t h e  s eco n d  l ev e l  a n d  t h e t h i r d  l e v el .  T h e ap p r o ach  o ccu r s   w h e n  t h e   i n d u ct i o n  p r o ces s  o f  t h e r u l e a t  each  l ev el ,   u s i n g  t h m et h o d  o f  f eat u r e s el ect i o n  t y p e e m b e dde d   [ 19] .     T h e o v er al l  d i m e n s i o n al  r ed u ct i o n  ap p r o ach  ca n  b e e x p l ai n ed  b y  r ef er r i n g t o F i gu r e  1.  T h e  pr oc e s s   o f  d ia g n o s is   in  t h e   s y s te m  b e g in s   w ith  a  p r e d ic tio n  a t le v e l 1 .  I f  th e  p e r c e n ta g e  o f  t h e  o u tp u t r is k  o f  c o r o n a r y   h e a r t d is e a s e  is  lo w  a t t h e   le v e 1 ,   th e n   f u r th e r   d ia g n o s i s   is  n o t n e c e s s a r y .  I n   th e s e   c o n d itio n s ,   it  w a s   n o t   Evaluation Warning : The document was created with Spire.PDF for Python.
I nd o ne s i a n J  E l e c  E ng  &  C o m p  S c i     I SSN :   2502 - 4752       A  T i e r e d A ppr oac on D i m e ns i onal  R e d uc t i on P r oc e s s  f or  P r e di c t i on of  ( W i har t o W i har t o )   493   ne c e s s a r y  to  a ttr ib u te  e x a m i n a tio n  a t le v e l 2  a n d  le v e l 3 .  I t s h o w s  t h a t t h e r e  h a s  b e e n  a  d i m e n s io n a l r e d u c tio n   o f  al l  at t r i b u t es  at  l e v el  2  an d  l ev el  3 .  I f  t h e p er cen t ag e o u t p u t  i n  t h e l ev el  1   w as  h i g h  s o  t h e r i s k  o f  co r o n ar y   he a r t  d i s e a s e   w a s  hi g h,  t he n f ur t h er  d i ag n o s i s  i s  r eq u i r ed  at  t h e s eco n d  l ev el .  I f  t h e d i ag n o s i s  at  t h e s eco n d   le v e w a s  s u f f ic ie n t to   m a k e  a  d e c is io n  f o r  th e  c lin ic ia n ,  f u r th e r  d ia g n o s is  to  th e  le v e l 3   w a s  n o t n e c e s s a r y .   T h e   c e s s a tio n   o f   d ia g n o s is   a t   le v e 2 ,   in d ir e c d i m e n s io n   r e d u c tio n ,   ie   a ll  a ttr ib u te s  a t l e v e 3   d o   n o n e e d   e xa m i na t i o n.     T h s eco n d   ap p r o ach ,   n a m el y   d i m e n s i o n al   r ed u ct i o n   at   ea ch   l ev el ,   es p eci al l y   l e v el   2   an d   l ev el   3 .   D i m e n s i o n al  r ed u c t i o n  p r o ces s  at  b o t h  l ev el s  o cc u r s   s i m u l t an eo u s l y   w i t h  t h i n d u ct i o n  p r o ces s  r u l e.   R e f e r r i n g  to  th e  r e s u lts  o f  t h e   te s t s h o w n  in  T a b le  5 ,  a t  th e  f ir s t le v e l t h e r e  is  a  d im e n s io n a l r e d u c tio n ,  ie  o n e   a ttr ib u te ,  d u e  to   u n a c c o m m o d a te d  a ttr ib u te  f b s  in  th e  s c o r e  ta b le  in  t h e  F R S   s t u d y   [ 20] .  A th e   s e c o n d  le v e l,   a ll a lg o r ith m s   f o r  in d u c tio n  r u le  c a n  r e d u c e  s o m e  a ttr ib u te s .  T he  C A R T  a l go r i t h m  c a p r o vi d e  t he  hi ghe s t   d i m en s i o n al  r ed u ct i o n ,  co m p ar ed  t o  o t h er  al g o r i t h m s .  T h e C A R T  al g o r i t h m  i s  al s o  cap ab l e o f  d el i v er i n g   pe r f or m a n c e  f or  t h e   h i gh e s t  pos t - t e s t  pr oba bi l i t y  pa r a m e t e r s  c o m pa r e d t o C 4. 5 a n d F D T ,  i e  b y   u s i ng  o n l y  on e   cp   a ttr ib u te   a le v e l   2 .   I n   p o p u la tio n - b as ed   p er f o r m a n ce  p ar a m et er s ,   t h p er f o r m an ce  o f   C 4 . 5  an d  F D T  i s   r e la tiv e l y  b e tte r  th a n  th e   C A R T .     A t t h e  le v e l 3 ,  th e  u s e  o f  C 4 . 5 ,  C A R T ,  a n d  F D T  a lg o r ith m s  is   n o t c a p a b le  o f  r e d u c in g  a ttr ib u te s .   T h e ab s en ce o f  r ed u ced  at tr ib u te s  i n d ic a te s   th a t  b o th  a ttr ib u te s  a t t h e  th ir d  le v e l   h a v e  a   h ig h  c o n tr ib u tio n   to   t h e r es u l t  o f  t h e d i ag n o s i s ,  s o  t h at  i t  can  n o t  b e r ed u ced .  T h e r es u l t i n g  p er f o r m a n ce at  t h e 3 r d  l ev el ,  f o r   popu l a t i on - b as ed  p er f o r m a n ce  p ar am et er s  i s  t h e C 4 . 5  i n d u ct i o n   a l go r i t h m  p r o vi d i n g t he  hi ghe s t  p e r f o r m a nc e ,   w h ile   f o r  in d iv id u a l - b a s e d  i s  a  C A R T .  D i m e n s io n a l r e d u c tio n  r e s u lts  o f  t h e  th r e e  a lg o r ith m s ,  C A R T   a lg o r ith m  a b le  to  r e d u c e  th e  a t tr ib u te  in  t h e  n u m b e r  o f  th e   m o s t.  R e f e r r i n g  to  t h e  n u m b e r  o f  a ttr ib u te s  th a t c a n   be  r e du c e d,  pos t - te ts  a n d  p r e - te s t p r o b a b ilit y ,  a n d   A U C ,  th e   F R S  +  C A R T - b as ed  f u zz y  r u l e b as ed   co m b i n at i o n s  ca n  p r o v i d e b et t er  p er f o r m a n ce t h a n  F R S  + C 4 . 5  an d  F R S  + F D T .     T h e  pr o pos e d di m e n s i on r e duc t i on   m ode l  i s  a l s o c a pa bl e  of  r e du c i ng  c h e c k i n g c os t s .  T h e  a b ilit y  i s   d i v i d ed  i n t o  t w o  ap p r o ach es .  F i r s t ,  t h e o v er al l  ap p r o ach  t o  t h e s y s t e m  i s  d y n a m i c.  S eco n d ,  co s t  r ed u ct i o n  at   ev er y  l e v el ,  es p eci al l y   t h e s ec o n d  an d  t h i r d  l ev el .  F es h k i  an d  S h i j an i   [2 ]   cl as s i f y  t h e r eq u i r ed  i n s p ect i o n  f ee s   f o r  e a c h  a ttr ib u te  i n to  4 ,  in  o r d e r  f r o m  t h e  lo w - c o s t  t o  t he  e xp e ns i ve .  G r o up i n g t he s e  a t t r i b ut e s  c a n b e  s ho w n   a s f o l l o w s:       1.   s e x,  a ge   2.   ch o l ,  r es t b p s ,  f b s ,  r es t ecg   3.   cp ,  t h al ac,  ex an g ,  o l d p eak ,  s l o p e.   4.   th a l,  c a   R ef er r i n g  t o  t h e cl as s i f i cat i o n   o f  i n s p ect i o n   f ees ,  t h e co s t  o f   t h e ex a m i n at i o n  l ev e l  1 <l ev el   2 <l ev el  3 .   I n  th e  p r o p o s e d  s y s te m   m o d e l ,  if  t h e  e x a m i n a tio n  o f  a  p a tie n t a t le v e l 1   w a s  s ta te d  t h a t t h e  r is k  o f  c o r o n a r h ear t  d i s eas e i s  l o w ,  t h er e i s  n o  n eed  f o r  ex a m i n a t i o n  at  l e v e l  2  an d  l ev el  3 .  D o  n o t  d o  t h ex a m i n at i o n  at  t h e   l ev el  2  a n d  l e v el  3 ,   m a k i n g   t h e co s t  t o  b e r ed u ced .  W h er e as ,  i f  at  t h f i r s t  s t ag e  i t  i s   s t at ed  t h at   i t   n eed s  a  f ur t he r  d i a g no s i s ,  i t  c a n b e   ex a m i n ed  at  l ev el  2 ,  an d  i f  at  t h at  l ev el  i s  e n o u g h  t o   m a k e a co n cl u s i o n ,  t h en  t h e   l ev el  3  ex a m i n at i o n   w a s  n o t   n eed ed  an y m o r e.     T h s eco n d   co s t   r ed u ct i o n   co n cep t ,   w a s   t h co s t   r ed u ct i o n   t h at   o ccu r s   a t   each   l e v el ,   es p eci al l y   at   le v e l 2  a n d  le v e l 3 .  T h is  is   d u e t o  t h e s eco n d  an d  t h i r d  l ev el  t h er w as  a r ed u ct i o n  i n  t h n u m b er  o f  at t r i b u t es ,   r e s ul t i n g i n r e d uc e d  c o s t s .   F o r  e xa m p l e ,   w he us i n g t he  C A R T  r ul e  i nd uc t i o n a l go r i t h m ,  a t  t he   s e c o nd  l e ve l  i t   w a s  ab l e t o  r ed u ce t h e at t r i b u t es  o f  E C G  e x a m i n at i o n  at  r e s t   an d  ex er ci s e,  i e r es t ec g ,  e x an g ,  t h a l ac,  o l d p eak ,   a nd  s l o p e .  R e f e r r i n g t o  t he   gr o up i n gs  b y F e s h ki  a nd  S hi j a ni   [2 ] ,  th e s e  a ttr ib u te s  b e lo n g  to  t he   s e c o nd  a nd  t he   t h i r d  g r o u p s ,  o r  co s t  i n  r el at i v el y  ex p en s i v e cat eg o r i es .  T h e l ar g e n u m b er  o f  at t r i b u t e s  t h at  can  b e r ed u ced   b y   t he   C A R T  i nd uc t i o n a l go r i t h m  r ul e ,   m a ki ng  t he  t o t a l  c o s t  o f  e xa m i na t i o n a t  l e ve l  2  i s  r e l a t i ve l y c he a p ,   c om pa r e d t o u s i ng  C 4. 5   a n d  F D T  a lg o r ith m s .   T h e  di m e ns i on a l  r e du c t i on  w i t h  c os t  c ons i de r a t i on   h a s  a l s o be e n  don e  i n   s o m e  pr e v i ous  s t u di e s ,   on l y  i n  pr e v i o u s   s t u di e s  di d not  u s e  a  t i e r e d a ppr oa c h .  F e s hk i  a n d S h i j a n i   [2 ]   r es ear ch ,  u s i n g  p ar t i cl e s w ar m   o p tim iz a tio n   w it h   f it n e s s   f u n c tio n s ,  c o n s id e r  c o s t,  s o  a s  to  r e d u c e  t h e  t w o  c o s tl y  a ttr ib u te s ,  c a  a n d  t h a l.  T h e   s a m e i s  d o n e b y   A r j en ak i  et . al   [4 ] ,   o nl y  t ha t  t he   s t ud u s e s   a  ge ne t i c  a l go r i t h m .   B o t s t ud i e s  ha ve  s o m e   d i s ad v an t a g es ,  n a m el y  t h u s e  o f  cl as s i f i cat i o n  al g o r i t h m s   w i t h  t h e b l ac k - box a ppr oa c h ,   m a k i ng   i t  di f f i c u l t  t und e r s t a n d t h e  pr oc e s s  of  di a g n os i s  b y  c l i n i c i a n s .  S e c on d,   do n ot  us e  a   t i e r e d a ppr oa c h ,  s o i n t h e  pr oc e s s  o f   d ia g n o s i s   m u s t c h e c k  a ll a ttr ib u te s  o f  th e  r e d u c tio n  r e s u lts ,  b e f o r e  b e in g   u s e d  f o r  in p u t d ia g n o s is  s y s te m .   T h e r es u l t  o f  co m p ar i s o n  o f  s y s t e m   m o d el  p er f o r m an ce d i a g n o s i s  o f  co r o n ar y   h ear t  d i s ea s w i t h  t h e   t i er ed  an d   n o n - t i er ed  ap p r o ach  i n  d i m e n s i o n  r ed u ct i o n  p r o ces s  ca n  b s h o w n  i n  T ab l e 6 .  I n  T ab l e 6 ,  i t  s h o w s   t h at  t h e p er f o r m an ce r e s u l t e d  f r o m  t h e s y s t e m   m o d el   w i t h  t h e t i er ed  ap p r o ach  b y   u s i n g  t h e b l ac k - box  c la s s i f ic a tio n  a l g o r ith m  b e tte r  th a n  t h e   tie r e d   m o d e l.  T h is   is  in d ic a te d  b y  t h e  p a r a m e te r s  o f  p o p u la tio n - b as ed   p er f o r m a n ce,  i e,  acc u r ac y  a n d  A U C   > 80% ,  on  t h e  N B  a n S V M  a l g or i t hm s .  M ode l  of  t he  di a g n os t i c  s y s t e m   w it h   th e  tie r e d  a p p r o a c h  is  b e tte r  th a n   t i e r e d b y  us i ng bl a c k - box c l a s s i f i c a t i on a l gor i t hm ,  t h a t  i s   f or   Evaluation Warning : The document was created with Spire.PDF for Python.
                      I SSN :   25 02 - 4752   I nd o ne s i a n J  E l e c  E ng  &  C o m p  S c i ,   V o l.   11 , N o .   2 A ug us t  2018   :   4 87     495   494   i nd i vi d ua l - b as ed  p er f o r m a n ce  p ar am et er s ,  t h a t  i s ,  t h m e an  v al u e o f  p o s t - t e s t  pr oba bi l i t y  a n d pr e - te s t   pr oba bi l i t y   > 43, 80% .  I n  t h e  n on - t i er ed   m o d el ,  t h e p ar a m e t er  v al u e i s  3 5 . 0 2 % ,  s o  t h er e i s   a d i f f er en ce o f   8 . 7 8 % .  I t  s h o w s  a b et t er  t i er e d  m o d el   w h en   u s ed   f o r  t h e d i ag n o s i s  o f  eac h  i n d i v i d u al .  T h e s a m e i s  t r u f o r   no n - b l ack - box  c l a s s i f i c a t i on  a l g or i t hm s ,   m or e  t i e r e d a pp r oa c h e s ,  bot h  f or  popu l a t i on - ba s e d a n d i n di v i du a l - b as ed  p er f o r m a n ce p ar a m et e rs .       T ab l 6 .  C o m p ar i s o n  o f  t i er ed  s y s t e m  p er f o r m an ce  w i t h   n o  t i er ed   A l g o r ith m   T h e  a l g or i t h m   a p p r oa c h   T i er ed   A ccu r acy  ( % )   AU C  ( % )   D i ffe re n c e s  o   p o s t  &  p r e - t es t   P r o b a b il it y  ( % )   NB   B l ack - b o x   -   8 3 , 89   8 3 , 65   3 7 , 78   S VM   B l ack - b o x   -   8 2 , 55   8 2 , 21   3 7 , 29   M LP   B l ack - b o x   -   7 7 , 52   7 7 , 32   2 9 , 99   C 4. 5   N on - B l ack - b o x   -   7 8 , 19   7 7 , 79   3 2 , 43   CA RT   N on - B l ack - b o x   -   7 8 , 52   7 8 , 06   3 3 , 53   FD T   N on - B l ack - b o x   -   7 5 , 84   7 5 , 41   2 9 , 88   C 4. 5+ F I S   N on - B l ack - b o x     8 1 , 21   8 0 , 43   4 4 , 76   CA RT + F I S   N o n  B l a ck - b o x     7 9 , 53   7 8 , 59   4 6 , 16   F D T + F IS   N on - B l ack - b o x     8 0 , 20   7 9 , 37   4 4 , 52       4.   CO NCL U S I O N   T h e t i er ed  d i m en s i o n  r ed u c t i o n   m o d el  i n  t h e p r ed i ct ed  s y s t e m  o f  co r o n ar y   h ear t  d i s eas w as  cap ab l of  pr ov i di n g  t w o di m e n s i o n a l  r e du c t i o n  p r o ces s es  at  o n ce.   T h e d i m e n s i o n al  r ed u ct i o n   t h at  o ccu r s  at   each   l ev el  an d  d i m e n s i o n  r ed u ct i o n  i s  d y n a m i cal l y   w h e n  u s e d  f o r  t h e d i ag n o s i s  o f  co r o n ar y  h ear t  d i s eas e.   D i m e n s i o n al   r ed u c t i o n   a t   eac h   l e v el ,   e s p eci al l y   t h s eco n d   an d   t h t h i r d   l e v el ,   F R S +C A R T   a lg o r ith m  c a n   p r o vi d e   t he   hi ghe s t  d i m e ns i o n r e d uc t i o n a nd  t he  b e s t  i nd i vi d ua l - b as ed  p er f o r m an ce.  T h e b es t  p o p u l at i o n - b a s e d  p e r f o r m a n c e   w a s  t h e  F R S + C 4 . 5  a lg o r ith m ,  b u t th e  r e s u lt in g  d i m e n s io n a l r e d u c tio n   w a s  lo w e s t.       R EF ER EN C ES   [ 1]   Li u  L  e t a l.   A  H y br i d C l a s s i f i c a t i on  S y s t e m   f or  H e a r t  D i s e a s e  D i a g nos i s  B a s e d on t he  R F R S   M e t ho d.   C om put .   M at h.  M e t h ods  M e d.   20 17:  1 11 .   [ 2]   F e s h k i M G  a n d  S h ija n i O S .   I m pr ov i ng  t he  H e ar t  D i s e as e  D i a gn os i s  by  E v ol u t i o nar y  A l gor i t hm  of   P SO  a nd F e e d   F or w ar d  N e ur al  N e t w or k A r tif ic ia l I n te llig e n c e  a n d  R o b o tic s  ( I R A N O P EN ) ,  Q a z vi n ,  I r a n,  20 16 :   48 53 .   [ 3]   A h m ad i  E ,  W eck m an  G D ,  an d   M as el  D T .  D eci s i o n  m ak i n g   m o d el  t o  p r ed i ct   p r es en ce o f  co r o n ar y  ar t er y  d i s e as us i ng  ne ur a l  ne t w or k  a nd C 5 . de c i s i on  t r ee.   J .   A m bi e nt  I nt e l l .  H u m ani z .  C om p ut .   20 17:  1 - 13.   [ 4]   A r j e na k i  H G ,  S ha hr a k i  M H N ,  a n d N our a f z a  N .   A  l ow  c os t   m ode l  f or  di a g nos i ng  c or ona r y  a r t e r y  di s e a s e  ba s e d on   ef f ect i v e f e at u r es .   I nt .  J .   E l e c t r on .  C om m u n.  C om pu t .  E ng .   20 15 ;  6( 1 ) :   93 9 7.   [ 5]   W i ha r t o W ,  K us na n to  H ,  a n d  H e r ia n to  H .  H y b r id  S y s te m  o f   T ie r e d  M u lt iv a r ia te  A n a l y s is  a n d  A r tif ic ia l N e u r a Ne t w o r k   f o r  C o r o n a r y  He a r t  Di se a se  Di a g n o si s.   I n t . J E l e c t r . C o m p u t .  E n g . I J E C E .  2 01 7;   7( 2) :  1 02 3 10 31 .   [ 6]   W i ha r t o W ,  H e r i a nt o H ,  a nd K us na nt o H .  T he  A na l y s i s  o f  P e r f o r m ace M o d el  T i er ed   A r t i f i ci al  N e u r al  N et w o r k   f o r   A s s es s m en t  o f  C o r o n ar y  H ear t  D i s eas e.   I n t . J . E l e c t r . C o m p u t . E n g .  I J E C E .  2 01 7;   7( 4) :  2 18 3 21 91.   [ 7]   W i ha r t o W ,  K us na nt o H ,  a n d H e r i a nt o H .  T i e r e d M o de l  B a s e d O n F uz z y  I nf e r e nc e  S y s t e m  F or  T he  D i a g nos i s  of   Co ro n ar y  H ear t  D i s eas e.   F ar  E as t  J .  E l e c t r on.  C om m u n.   20 16;  1 6( 4 ) :  98 5 1 00 0.   [ 8]   H id a y a t R  a n d   P r im a s a r i I .  M e to d o lo g P e n e litia n   P s ik o d ia g n o s t ik a .   Bu l.  Ps ik o l.   2 01 1;  19( 2) :   8 1 - 92 .   [ 9]   D et r an o  R ,  J o n as i  A ,  S t ei n b r u n n   W ,  an d   P f i s t er er  M .  H ear t  D i s eas e D at as et .  U n v er s i t y   C a lif o r n ia  I r v in e ,  C a lif o r n ia ,   198 8.   [ 1 0]   W i l s on  P W F ,  D A g os t i no R B ,   L e vy  D ,  B e l a ng e r   A M ,  S i l be r s h a t z  H ,  a nd K a n ne l  W B .   P r e d i c t i on  of  C or o na r y   H ear t  D i s eas e U s i n g  R i s k  F act o r  C at eg o r i es .   C ir c u la tio n .  19 98 ;   9 7 ( 18) :  18 37 18 47 .   [ 1 1]   R a m en t o l  E ,  C ab al l er o  Y ,  B el l o  R ,  a n d  H er r er a F .  S M O T E - R S B   * :  a  hy br i d pr e pr oc e s s i ng  a p p r oa c h ba s e d  o ov e r s a m pl i ng  a nd unde r s a m pl i ng  f or  hi g h i m ba l a nc e d da t a - s e t s  us i ng  S M O T E  a nd r oug h s e t s  t he o r y .   Kn o wl .  In f .   S ys t .   2 01 2;  33( 2) :  24 5 265 .   [ 1 2]   H s s i na  B ,  M e r bo uha  A B ,  E z z i k our i  H ,  a nd  E r r i t a l i  M .   A  c o m pa r a t i v e  s t udy  of  de c i s i on t r e e  I D 3 a nd C 4 . 5.   I n t .  J.   A dv .  C om put .  Sc i .  A p pl .  I J A C SA .  201 4;  4( 2) :   1 3 1 9.   [ 1 3]   B a i ha qi  W M ,  S e t i a w a n N A ,  a nd A r di y a nt o I .   R ul e  e x t r ac t i o n f or  f uz z y  e x pe r t  s y s t e m  t o di ag nos e  C or on ar y  ar t e r y   d i se a se I n t er n at i o n al  C o n f er en ce o n   I nf or m a t i on T e c hnol og y ,  I nf or m a t i on S y s t e m s  a nd E l e c t r i c a l  E ng i ne e r i ng   (ICIT IS E E ) .  Y og y a k a r t a ,  I ndone s i a ,  20 16 :  1 36 1 41.   [ 1 4]   S u J  a nd Z ha ng  H .  A  f a s t  de c i s i on t r e e  l e a r ni ng  a l g or i t hm .   T he  21 s t  nat i on al  c o nf e r e nc e  on A r t i f i c i al  i nt e l l i ge nc e B os t on ,  M a s s a c hus e t t s ,   20 06,  1:  5 00 50 5.   [ 1 5]   B r ei m an  L ,  F r i ed m an  J ,  S t o n e C J ,  an d  O l s h en  R A .   C l as s i f i c at i on   and  R e gr e s s i o n T r e e .  N e w  Y or k :  C ha pm a n A nd  H a l l ,  1 98 4.   Evaluation Warning : The document was created with Spire.PDF for Python.
I nd o ne s i a n J  E l e c  E ng  &  C o m p  S c i     I SSN :   2502 - 4752       A  T i e r e d A ppr oac on D i m e ns i onal  R e d uc t i on P r oc e s s  f or  P r e di c t i on of  ( W i har t o W i har t o )   495   [ 1 6]   K i m  J ,   L e e J ,   an d  L e e Y .   D at a - M i ni ng - B a s e d C or ona r y   H e a r t   D i s e a s e  R i s k P r e di c t i on M ode l  U s i ng  F u zz y  L o g i an d  D eci s i o n  T r ee.   H e al t hc .  I nf or m .  R e s .   2 01 5;  2 1( 3) :  1 67 - 17 4.   [ 1 7]   E z hi l m a r a n D  a nd J os e ph   P R B .   F u z z y i n f er en ce s ys t em  f o r  f i n g er  vei n  b i o m et r i c i m a g es I nt e r na t i ona l  C onf e r e nc e   on I nv e nt i v e  S y s t e m s  a nd C on t r ol  ( I C I S C ) .  I ndi a ,   20 17 :  1 4.   [ 1 8]   P o k h ar el  D   e t a l.   Es tim a tio n  o f  1 0 - y ear  r i s k  o f  co r o n ar y  h ear t  d i s eas e  i n  n ep al es e p at i en t s  w i t h  t y p e  2  d i ab et es :   F r a m i ng ha m   v e r s us  U ni t e d K i ng d om  pr os pe c t i v e  di a be t e s  s t udy .   N or t h A m .  J .  M e d.   Sc i .   20 15;  7( 8) :  347 35 5.   [ 1 9]   J e ns e n R  a nd  S he n  Q .   C om put at i on al  I nt e l l i ge nc e   an F e at ur e   S e l e c t i on  :   R oug an F uz z y   A pp r oac he s .  U S A:   I E E E  P r e s s 2 0 0 8 .   [ 2 0]   S c hi s t e r m a n E F  a nd W hi t c om b B W .  C or ona r y  a ge  a s  a  r i s k   f a c t or  i n  t he  m odi f i e d F r a m i ng ha m  r i s k  s c or e .   B M M e d.  I m agi ng .  2 00 4;   4( 1) :  1 - 9.     Evaluation Warning : The document was created with Spire.PDF for Python.