I am looking for ways to improve my average silhouette width for my data.
So far I have removed columns from the dataframe that have NA values, columns with many outliers as well as columns that are calculations based on other columns.
I have tried to remove the outliers from the other data columns but this makes the average silhouette worse.
Is there any way I can increase my average silhouette width more?
Dear Shelby,
Please, clarify what you mean by 'columns' - objects or variables?
Other questions arise, too: why do you want to increase average silhouette? How did you produce the classification? Do you want to modify your classification to fit your data better, or do you prefer to change you data to fit the classification?
I have two publications on silhouette width, perhaps you find them helpful:
https://www.biorxiv.org/content/10.1101/630384v2.abstract
Article Silhouette width using generalized mean—A flexible method fo...
Best wishes,
Attila
- Badges
- Science topic