A Bimax Biclustering Analysis of Crime Types by Nationality in Türkiye

Creative Commons License

Kocaturk A., Albasar D., ÖRKCÜ H. H.

JOURNAL OF POLYTECHNIC-POLITEKNIK DERGISI, cilt.29, sa.4, 2026 (ESCI)

Özet

This study examines structural relations between nationalities and crime types in T & uuml;rkiye using the Bimax biclustering algorithm. Official judicial records from 2019-2023 (with varying nationality and crime type structures by year) were converted to binary format and analyzed in the R environment. Bimax identified compact biclusters that uncover recurring co-occurrence patterns across nationalities over the years-for example, the joint appearance of assault, qualified theft, and drug-related offenses-and highlight overlaps between groups rather than single-variable frequencies. The workflow comprises data binarization, biclustering with minimum row/column constraints, and heat-map visualization to interpret membership. Methodologically, the paper demonstrates the utility of Bimax for sparse, high-dimensional administrative data, offering an interpretable alternative to province-level analyses. Substantively, the findings provide policy-relevant signals for targeted prevention and integration strategies, indicating that crime profiles among foreign nationals are shaped by migration-related and socio-economic factors. Because the approach can produce consistent results under different threshold values and algorithm settings, it can be easily adapted to similar administrative datasets and different contextual analyses.