Değiştirilebilir konum süresine sahip takip cihazlarında kümeleme parametrelerinin tahmini ve anomali tespiti

Abstract

In this study, it was aimed to detect anomalies in the location behavior of objects followed by a tracking device.Theory and Methods: ST-DBSCAN (Spatial-Temporal Density-Based Spatial Clustering of Applications with Noise) density-based clustering algorithm was applied on the data obtained, and weekly patterns were determined for the subject to be located at which time intervals. The input parameters of the ST-DBSCAN algorithm vary according to the frequency of the data from the tracker and the total number of data packets. In this context, the parameters used in the St-DBSCAN algorithm, as well as the frequency of sending data and the number of data packets, are labeled according to the behavior of the object being followed. On these tagged data, linear regression and artificial neural networks methods were compared and a model was proposed that could predict clustering parameters. Results:Weekly patterns were determined by methods developed using information about the object being followed, and these patterns were considered to be normal behaviors of the object being tracked. The instantaneous position is defined as an anomaly if the data obtained is contrary to the pattern. Conclusion:In this study, the temporal data of the object, called time and coordinate data, were grouped with similar ones with the help of clustering algorithms, and were defined as anomalies when data other than sets that were normally considered to be able to predict with these sets.

Akan veri kümeleme probleminde ağaç veri yapılarının performans karşılaştırması

“…Makine öğrenmesinin kapsadığı algoritmalar dinamik yapıda olup, mevcut veriye göre modeller oluşturmayı amaçlamaktadır [4]. Kümeleme, makine öğrenmesi kapsamında çalışılan problemlerden birisi olup, veri kümesi içerisindeki verilerin benzerliklerine göre gruplandırılmasını sağlamaktadır [5]. Akan veri kümeleme, bir sisteme sürekli olarak gelen verilerden anlamlı bilgiler çıkarmayı amaçlayan makine öğrenmesinin önemli konularından biridir.…”