Credit card fraud detection by dynamic incremental semi-supervised fuzzy clustering
Gabriella Casalino, Giovanna Castellano, Corrado Mencar · 2019
The problem of credit card fraud detection is approached by a semi-supervised classification task on a data stream.The DISSFCM algorithm is applied, which is based on Dynamic Incremental Semi-Supervised Fuzzy C-Means that processes data grouped in small-size chunks.Experimental results on a real-world dataset of credit card transactions show that DISSFCM has comparable results with some fully-supervised stream data classification methods, also in presence of a high percentage of unlabeled data.