Open AccessTarget backcloth, series length, and the accuracy of geographic profiling. Simulation analysis of target backcloth, series length, and the accuracy of geographic profiling
Author(s) -
Michał Górski
Publication year - 2022
Publication title -
z zagadnień nauk sądowych/problems of forensic sciences
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.119
H-Index - 12
eISSN - 2720-5983
pISSN - 1230-7483
DOI - 10.4467/12307483pfs.20.006.15446
Subject(s) - profiling (computer programming) , series (stratigraphy) , computer science , statistics , geographic information system , data mining , algorithm , mathematics , geography , cartography , geology , paleontology , operating system
The main aim of this research was to establish a minimum series length when crime opportunities are not distributed at random. In addition, the behaviour of two methods of assessing the accuracy of a geographic profiling algorithms were analysed. The second objective was to analyse the existence and properties of the buffer zone in the created model. For the purpose of this research, a simulation was used. Data showed significantly different behaviour in the two tested methods of assessing the accuracy of geographic profiling, with hit score percentage being the significantly more sensitive measurement. The tests allowed all of the effects attributed to the buffer zone to be obtained, despite the fact that it was not included in the model itself. For series as long as nine offences accuracy in non-random conditions was similar to series length five and uniform distribution of opportunities.