6.3 Result analysis and
Discussion
\label{result-analysis-and-discussion}
This subsection describes the results of our study. Table 4 illustrates
the confusion matrix obtained for dengue dataset using ANN, DT, NB, PSO
and PSO-ANN approaches. The confusion matrix is used to derive the
results of performance measures used in this study. This matrix contains
actual and predicted values of all data instances of dengue disease. The
experimental results of all above mentioned approaches are described in
Table 5. It is seen that PSO-ANN based approach provides better results
in comparison to other approaches being compared. It is also noticed
that PSO-ANN approach based diagnostic model obtains higher accuracy,
sensitivity and specificity rate. Further, it is also observed that DT
approach provides worst performance among all. It is also revealed that
the performance of ANN approach is significantly improved by optimizing
its weight and bias parameters through PSO method. Moreover, the ranking
of each approach is also computed using performance measures. Again, it
is observed that PSO-ANN approach obtains the first rank for all
parameters and DT approach have worst rank.
Table 3: Parameters setting of DT, ANN, NB and PSO approaches