Which statement is true regarding k-means clustering?

K-means clustering is a popular clustering method that can be sensitive to the arrangement of the data, especially in terms of how distances are calculated. A major drawback of k-means is its sensitivity to outliers and noise, which can distort the formation of clusters. When data points are inversely related (i.e., they influence the mean in a way that doesn't represent the underlying groups well), it can lead to misleading results. This issue is particularly pronounced when clusters are not spherical or when their sizes vary significantly, which can happen often in real-world data. Thus, inversions can lead to misrepresentation of cluster centers, making them a significant drawback in the application of the k-means algorithm.

While some clustering methods can handle categorical variables better, k-means fundamentally relies on calculating distances, which is not directly applicable to non-numeric types. This further highlights that k-means clustering indeed has limitations, such as susceptibility to inversions, making this statement true.