id	author	title	date	pages	extension	mime	words	sentence	flesch	summary	cache	txt
american_scientific_journal-9700	Huseyin Carpanali ; Ayse Humeyra Bilge ; Arif Selcuk Ogrenci ; Tarkan Ozmen; Ayse Tosun ; Kubra Cakar	Detection of Rare Events: Cluster Based Preprocessing of the Training Set: The Case on Complaints for Invoice Time Series	2024	15	.pdf	application/pdf	6109	304	60	The method we propose is applied to the case of the detection of customer complaints to monthly invoices of a major telecommunication company. The mean, relative standard deviation and relative range, as basic features of the clustering algorithm of the reference [10], are proved to be useful for the analysis of customer complaints, but for the purposes of preprocessing the training set for better performance of machine learning algorithms, we are rather interested in the variability of the items in each cluster.	cache/american_scientific_journal-9700.pdf	txt/american_scientific_journal-9700.txt