Computer scienceData scienceMachine learningClusteringClustering evaluation

Introduction to clustering evaluation

The straightforward approach

Report a typo

In the external evaluation section, we mentioned how one might just look at the count of correctly clustered samples (this idea was introduced as the Rand index or accuracy). Why using accuracy as a performance metric might not be enough for a correct assessment?

Select one or more options from the list

Accuracy can't distinguish different types of errors

Because the ground truth labels are required

Accuracy can't deal with imbalanced data

Because accuracy doesn't have a fixed score range

___

Create a free account to access the full topic