In internal performance evaluation, clustering is evaluated based on feature data only. This method does not use any target label information. These evaluation measures assign better scores to clustering methods that generate well-separated clusters. Here, a high score does not guarantee effective clustering results.
Internal performance evaluation helps us to compare multiple clustering algorithms but it does not mean that a better-scoring algorithm will generate better results than other algorithms. The following internal performance evaluation measures can be utilized to estimate the quality of generated clusters:
