AI Engineering Degree Practice Test - Prep, Questions & Study Guide

Measuring the performance of a k-means clustering model without ground truth can effectively be accomplished by evaluating the average distance between data points and their corresponding cluster centroids. This method provides a clear sense of how closely related the data points are to their clusters: the smaller the distances, the more tightly grouped the points are within each cluster, indicating better clustering performance. Using average distances effectively quantifies the compactness of the clusters, which is a crucial factor in assessing their quality.

In contrast, simply counting the number of clusters formed does not provide direct insight into the quality or cohesiveness of those clusters. Similarly, assessing the total number of features does not measure performance regarding how well the clustering reflects the underlying structure of the data. Finally, calculating the time taken for clustering might inform about the efficiency of the algorithm but does not provide insights into the effectiveness or accuracy of the clustering results themselves. Therefore, measuring the average distance from points to centroids remains the most relevant approach in this context.