Control Systems and Computers, N2, 2018, Article 9
DOI: https://doi.org/10.15407/usim.2018.02.087
Upr. sist. maš., 2018, Issue 2 (274), pp. 87-96.
UDK 004.942
Natalia Kuznietsova, PhD in Technical Sciences, Associate Professor, natalia-kpi@ukr.net
Petro Biduyk, Doctor of Technical Sciences, Professor,
pbidyuke_00@ukr.net.
Mathematical Methods of System Analysis department of the Institute for Applied Systems, Analysis NTUU «Igor Sikorsky KPI», Peremohy ave., 37, Kyiv, 03056, Ukraine
Identification of the fraud risk during
tender pURCHASES by survival theory methods
Introduction. Transparency in public procurement is a vivid information for foreign investors and the international community, which ensures an equal access and opportunities for all companies in the market, indicating a lack of corruption and self-regulation of the market. For Ukraine, where a special platform for open online purchases was created, the task of verifying and detecting conspiracies or opaque contracts is rather relevant, as media are increasingly accused of opacity and conspiracy in conducting such tenders.
The purpose of the article is to develop the new models for the company’s identification and classification in order to identify unfair competition and conspiracy.
Methods. Standard methods of data showed rather high error (at 30% level), and therefore it was proposed to construct the survival models based on the proportional Cox risks and Kaplan-Mayer estimates.
Results. Survival functions and risk functions allowed us to determine behavior of fraud suspected companies in time, as well as depend on the time spent on the platform, the number of applications submitted and rejected on the tenders. Separately, there were built survival functions for the companies in the number of rejected tenders, and a tendency for a longer stay on the platform of companies that had certain arrangements with each other and participated in tenders was determined.
Conclusions. The proposed approach to the analysis and classification of companies on the basis of the survival models showed high results of the suspicious companies detection and significantly higher probability of survival and subsequent work in time compared with the companies real participants of the bidding. Higher probabilities of companies’ survival are also observed if they have a large number of cases in bidding, and have repeatedly disqualified from bidding. This is a confirmation that the models correctly identify the suspicious companies, which, unfortunately, are present and remain active on the Ukrainian tender market.
Introduction. Transparency in public procurement is a vivid information for foreign investors and the international community, which ensures equal access and equal opportunities for all companies in the market, indicating a lack of corruption and self-regulation of the market. For Ukraine, where a special platform for open online purchases was created, the task of verifying and detecting conspiracies or opaque contracts is rather relevant, as media are increasingly accused of opacity and conspiracy in conducting such tenders.
The purpose of the article is to develop new models for the identification and classification of the companies which are participating in tenders in order to identify unfair competition and conspiracy.
Methods: Standard methods of data showed rather high error (at 30% level), and therefore it was proposed to construct survival models based on proportional Cox risks and Kaplan–Mayer estimates.
Results: Survival functions and risk functions allowed us to determine behavior of fraud suspected companies in time, as well as depend on the time spent on the platform, the number of applications submitted and rejected on tenders. Separately, there were built survival functions for companies in the number of rejected tenders, and a tendency for a longer stay on the platform of companies that had certain arrangements with each other and participated in tenders was determined.
Conclusions: The proposed approach to the analysis and classification of companies on the basis of survival models showed high results of the detection of suspicious companies and significantly higher probability of survival and subsequent work in time compared with companies real participants of the bidding. Higher probabilities of survival have also been seen for companies that have had a large number of cases in bidding, and have been repeatedly disqualified from bidding. This is a confirmation that the models correctly identify suspicious companies, which, unfortunately, are present and remain active on the Ukrainian tender market.
Download full text! (In Russian).
Keywords: tender purchases, fraud risks, Cox proportional risk model, Kaplan—Mayer estimation.
REFERENCES
1. Zakon Ukrainy Pro Publichni Zakupivli. [online] Available at: <http://zakon2.rada.gov.ua/laws/show/922> [Accessed 24 Jan. 2016]. (In Ukrainian).
2. ProZorro: publichni Zakupivli. [online] Available at: < https://prozorro.gov.ua> [Accessed 12 Dec. 2017]. (In Ukrainian).
3. ZAYCHENKO, YU.P., 2006. Osnovi proektuvannia intelektualnih system, Kyiv, Slovo. 352 p. , (In Ukrainian).
4. CHUBUKOVA, I.A., 2008. Data Mining, Moskow. Binom LBZ, 384 p. (In Russian).
5. Kuznetsova, N.V., Bidyuk, P.I. Modeling of credit risks on the basis of the theory of survival. Journal of Automation and Information Sciences, Volume 49, Issue 11, 2017, P. 11–24.
https://doi.org/10.1615/JAutomatInfScien.v49.i11.30
6. Bidyuk, P.I., Romanenko, V.D., Tymoshuk, O.L., 2013. Time series analysis [Analiz chasovih riadiv], Kyiv, Polytechnika, 600 p. (In Ukrainian).
7. Cox, D.R., Society, S.B., 2007. “Regression models and life-tables. Methodological”, 34 (2). pp. 187–220.
8. CAO, R., VILAR, J.M., DEVIA, A., 2009. “Modelling consumer credit risk via survival analysis”. SORT 33 (1), January-June 2009, pp. 3–30.
9. ALLISON, P.D., 2010. Survival Analysis Using SAS: A Practical Guide. Cary, NC: SAS Institute Inc., 324 p.
Received 16.04.2018
