Page 204 - Contributed Paper Session (CPS) - Volume 4

CPS2174 Septian R. et al.
bigger. In other hand, the standard deviation for bigger size of tends to
increase. It can be said that smaller size of in this case can make the
prediction very accurate.
Table 1 Mean and standar deviation of evaluation criteria in each

Evaluation
Criteria Value m=50 m=100 m=150
Accuracy mean 0.998 0.998 0.994
sd 0.007 0.004 0.012
Sensitifity mean 0.999 0.998 0.992
sd 0.006 0.008 0.020
Specificity mean 0.997 0.998 0.996
sd 0.010 0.005 0.009

In addition, from 100 replications also can be investigated that there are
many iteration in the replications can predict the testing data very well in
perfect condition that discribed in Figure 2.

100
87
82
80 69
Replication 60

40

20

0
m = 50 m = 100 m = 150
Number of predictor variables in model candidate ()

Figure 2 Bar chart of number of replication which has very good prediction
based on
Therefore, when in the model candidate use = 50 , 87% of 100
replications has very good performace of prediction. Besides when in the
model candidate use = 100, and = 150. Because of that, in this case it
can be showed that by using = 50 can give the prediction performance very
good in accuracy, sensitifity, and specificity.

193 | I S I W S C 2 0 1 9