25. Fig. 10 shows the execution time for rule generation and cluster formation for diabetic and heart disease data set.
For two data sets, the cluster formation time is less than compared to the rule generation. The rule generation takes
more time because it converts all the original data set into low, medium, high value to generate the candidate rules.
26. Fig 15 and 16 shows the purity and NMI result for diabetic and heart disease data set. For diabetic data set, the
purity achieved 77%, and for heart disease 81%. The NMI value is more than 70% for both diabetic and heart disease
data set when increasing the number of rules.
27. Fig 19 shows the accuracy comparison of 3 different algorithms. Compared to ANN-ICA (Integrated Component
Analysis) and LNF-PCA, the proposed algorithm obtains higher accuracy.