Forgy
Forgy is a surname that appears in multiple areas of computer science. In clustering, the Forgy method, also called the Forgy initialization, is a simple technique used to initialize the k-means algorithm. In this method, k observations are chosen at random from the data set and used as the initial centroids. The approach is fast and easy to implement, since the initial centroids are actual data points. However, because the selection is random, the method can lead to poor convergence if the chosen points do not reflect the data’s underlying structure, and it is sensitive to outliers. In practice, practitioners may run the algorithm multiple times or use more sophisticated initializations to improve robustness.
The name is also associated with Charles Forgy, a computer scientist known for developing the CLIPS expert
In summary, Forgy can refer to the statistical initialization method used in k-means clustering as well as