raw

  ----------------------------------------
jrip: 68.75
(P5 >= 3) => CLASS=Confused (8.0/2.0)
(E5 >= 1) => CLASS=Confused (6.0/2.0)
 => CLASS=Not (18.0/0.0)
 
   a  b   <-- classified as
 18  4 |  a = Not
  6  4 |  b = Confused
  
  ----------------------------------------
  naive baues: 81.25
  
  === Confusion Matrix ===

  a  b   <-- classified as
 18  4 |  a = Not
  2  8 |  b = Confused
  
    ----------------------------------------
    
    discretize attributes into 3 bins, struggle to do equal size
    
    jrip: 75%
    
  
  (r1 = '(1-2.5]') => CLASS=Confused (5.0/0.0)
 => CLASS=Not (27.0/5.0)
 
  a  b   <-- classified as
 20  2 |  a = Not
  6  4 |  b = Confused
     
      ----------------------------------------
      
 bayes: 
 
 acc: 78%
  a  b   <-- classified as
 18  4 |  a = Not
  3  7 |  b = Confused
          
        ----------------------------------------
        
  prism: 59%
  
  a  b   <-- classified as
 16  0 |  a = Not
  7  3 |  b = Confused
              
        
          ----------------------------------------
          
     feature subset selection on the discretized space
     
     wrapper. target: PRISM
     
         10(100 %)     1 r1 
           8( 80 %)    16 P6
           7( 70 %)    19 O3
           1( 10 %)    17 O1
           1( 10 %)    18 O2
           1( 10 %)    21 M2
           1( 10 %)     3 r3 
           1( 10 %)     6 E1
     
    prism on r1 p6 o3 
    Accuracy:   84.375  %      
        
  a  b   <-- classified as
 21  0 |  a = Not
  1  6 |  b = Confused  
          
    Prism rules
    
If O3 = '(0.5-1.5]' then Not
If r1 = '(-inf-1]'
   and P6 = '(-inf-0.5]' then Not
If P6 = '(0.5-2.5]'
   and r1 = '(2.5-inf)' then Not
If r1 = '(-inf-1]'
   and P6 = '(0.5-2.5]'
   and O3 = '(-inf-0.5]' then Not
If r1 = '(1-2.5]' then Confused
If P6 = '(2.5-inf)' then Confused
If r1 = '(2.5-inf)'
   and P6 = '(-inf-0.5]' then Confused
If O3 = '(1.5-inf)'
   and r1 = '(-inf-1]'
   and P6 = '(0.5-2.5]' then Confused
If P6 = '(0.5-2.5]'
   and r1 = '(-inf-1]'
   and O3 = '(-inf-0.5]' then Confused
   
 ----------------------------------------
PRISM: r1 r6

Accuracy: 90.63
  a  b   <-- classified as
 22  0 |  a = Not
  2  7 |  b = Confused

If r1 = '(-inf-1]'
   and P6 = '(-inf-0.5]' then Not
If P6 = '(0.5-2.5]'
   and r1 = '(2.5-inf)' then Not
If r1 = '(-inf-1]'
   and P6 = '(0.5-2.5]' then Not
If r1 = '(1-2.5]' then Confused
If P6 = '(2.5-inf)' then Confused
If r1 = '(2.5-inf)'
   and P6 = '(-inf-0.5]' then Confused
If P6 = '(0.5-2.5]'
   and r1 = '(-inf-1]' then Confused
   
   
   Prism: r1
   accuract: 84.37
   
  a  b   <-- classified as
 22  0 |  a = Not
  5  5 |  b = Confused

 ----------------------------------------
 
 FSS on naive bayes on discretized data
 
          10(100 %)     1 r1
           8( 80 %)    15 P5
           3( 30 %)    22 M3
           2( 20 %)    18 O2
           2( 20 %)     6 E1
           2( 20 %)    12 P2
           1( 10 %)    14 P4
           1( 10 %)    11 P1
           1( 10 %)    16 P6
           1( 10 %)    19 O3

bayes: r1 p5

accuracy =  87.5    %

  a  b   <-- classified as
 21  1 |  a = Not
  3  7 |  b = Confused
  
bayes: r1
accuracy: 78.13%

  a  b   <-- classified as
 20  2 |  a = Not
  5  5 |  b = Confused
   
 ----------------------------------------
 
FSS on jrip on discretized data

           6( 60 %)     1 r1
           4( 40 %)    15 P5
           2( 20 %)     2 r2
           1( 10 %)     6 E1
           1( 10 %)     7 E2
           1( 10 %)     8 E3
           1( 10 %)    11 P1
           1( 10 %)    12 P2
           1( 10 %)    13 P3
           1( 10 %)    18 O2


accuracy: 75 %
  a  b   <-- classified as
 20  2 |  a = Not
  6  4 |  b = Confused