Hello,
I am trying to analyze dietary patterns based on 40 food groups using kmeans cluster analysis on 5062 participants . I used the Calinski & Harabasz Pseudo-F statistic to decide on the best clustering option. 5 clusters had the highest Pseudo-F; however, when I conduct the cluster kmeans specifying k5, I get a different table every time I run an iteration using a different cluster name. Is it ok to also use the same Pseudo-F statistic when choosing between the different iterations for 5 clusters? Thank you
I am trying to analyze dietary patterns based on 40 food groups using kmeans cluster analysis on 5062 participants . I used the Calinski & Harabasz Pseudo-F statistic to decide on the best clustering option. 5 clusters had the highest Pseudo-F; however, when I conduct the cluster kmeans specifying k5, I get a different table every time I run an iteration using a different cluster name. Is it ok to also use the same Pseudo-F statistic when choosing between the different iterations for 5 clusters? Thank you
