Zgadzam się
Nasza strona zapisuje niewielkie pliki tekstowe, nazywane ciasteczkami (ang. cookies) na Twoim urządzeniu w celu lepszego dostosowania treści oraz dla celów statystycznych. Możesz wyłączyć możliwość ich zapisu, zmieniając ustawienia Twojej przeglądarki. Korzystanie z naszej strony bez zmiany ustawień oznacza zgodę na przechowywanie cookies w Twoim urządzeniu.
This study provides a comparison of the efficiency of anomaly detection in data using Isolation Forest (IF) combined with k-Means and Fuzzy C-Means algorithms. It also presents how to determine the anomaly score from the clustering results using the triangular and Gaussian membership functions. The number of clusters, the significance of the anomaly score obtained from the clustering process, and the degree of fuzziness of the clusters are additionally taken into account when testing the efficiency of anomaly detection. Moreover, we demonstrate that in most of the examined datasets, preceding IF with clustering algorithms allows obtaining significantly better results. Furthermore, combining IF with Fuzzy C-Means produces better results than combining it with k-Means. The results discussed in this paper allow one to decide which clustering method to use when combining it with IF to detect anomalies in the data. In addition, a comprehensive analysis presented in the paper sheds the light on the procedure of a choice of the parameters of the algorithms to get possibly the best results.
