The basic requirement for cluster analysis is to determine the number of clusters to be
formed by clustering algorithm. To achieve the solution for this, we used several information
criteria such as Akaike Information Criteria (AIC) [30] Bayesian Information
Criterion (BIC) [31] and Consistent AIC (CAIC) [32]. We generated 15 models for 1
cluster to 15 clusters. Figure 2 illustrates the evolution of BIC, AIC and CAIC for the 15
models generated. It shows that there is a reduction in the values of AIC, BIC and CAIC
with an increase in the number of clusters. Based on the Fig. 2 (a low score is considered
as good), we select the model with 6 clusters as there is no improvement after this. Our
selection also follows the approach used by previous studies [2, 24].