I am doing k means clustering analysis with my data and I know that one of the strategies to look at the stability of the clusters is Silhouette plot, taking that into account when evaluating the average Silhouette width I saw that it is judged taking into account a range from -1 to 1, being that the closer to 1 the better. However when looking at several articles I see that many of the averages are in the range of 0.2 to 0.6 and they are considered stable. I remember reading that average values above 0.2 is optimal, I wanted to confirm this information and if you could explain this to me.
According to my data from the elbow and silhouette graph the optimal number of clusters is 3 but when I make the average graph this is 0.23. Is this considered too low?
Thank you very much.
