Computer scienceData scienceMachine learningClusteringClustering evaluation

Introduction to clustering evaluation

Internal vs external measures

Report a typo

Select the statements that are true for internal and external clustering evaluation:

Select one or more options from the list

Internal metrics covered in the theory section make assumptions about the clusters, namely, they should be spherical and well-separated

Internal metrics covered in the theory section are suitable for the evaluation of non-centroid based approaches (such as DBSCAN)

External metrics don't make assumptions on the clustering structure

External metrics could be used to compare clusterings from different algorithms

External metrics won't show the true performance of centroid-based algorithms (such as K-Means)

External metrics rely on the presence of ground truth labels

___

Create a free account to access the full topic