By Andriy Krysovatyy, Hrystyna Lipyanina-Goncharenko, Svitlana Sachenko, Oksana Desyatnyuk

« Fictitious business – is the creation or acquisition of business entities in order to cover up illegal activities or activities that are prohibited. Investigation of economic crime takes a lot of time for law enforcement officers, so in this regard, the development of an algorithm for detecting a fictitious enterprise based on the classic method of machine learning, namely Support Vector Machine Classification, will develop a single software environment for rapid detection of economic crimes. To build the method, data from 1,100 companies operating in Ukraine were used. The data presented in in the set logical binary values are from 355 fictitious enterprises. Modeling of the Support Vector Machine was performed by 3 approaches: linear, polynomial and radial. The best results are obtained from classification by polynomial approach. The training sample showed evaluation results at 100%, and the test sample showed evaluation at 99.7%. Also, the confusion matrix showed quite good results. »

Cliquer pour accéder à paper46.pdf

%d blogueurs aiment cette page :