This paper explores the spatial distribution of crime incidences in Nigeria and evaluates the association between the geographical variations and the socio-demographic determinants of crimes. The analyses are based on 2017 reported crime Statistics obtained from the Nigeria‘s National Bureau of Statistics. This paper analysed the spatial patterns of four types of crimes (armed robbery, theft, rape and kidnapping) in relation to their geographical distributions across states in Nigeria. In contrast to the traditional regression analysis, a Poisson mixed model was formulated to incorporate the spatial dependence effects (clustering) and the specific state-level heterogeneity effects of crimes. The study modelled six explanatory variables (unemployment rate, population density, education index, Gross National Income (GNI), percentage males population (PMP), age 18–35 years and policing structure) as the determinants of crimes in Nigeria. A full Bayesian approach via Markov Chain Monte Carlo simulation was used to estimate the model parameters. The results show that the unemployment rate was positively associated with rape, kidnapping and armed robbery, but negatively associated with theft. The results further reveal that GNI and PMP show positive correlation with all the crimes. In addition to the risk factors of the crimes, the proportion variation attributed to clustering effect of the total variation was explained by 29.27 % in armed robbery incidents, 31.30% for theft (stealing), 27.07% for kidnapping and 41.40% in rape cases occurrence. Our approach also produces spatial predictive maps that identified areas of high crime concentration, which can assist the relevant agencies in crimes prevention, effective policing and areas needed urgent attention.

本文探讨了尼日利亚犯罪事件的空间分布，并评估了犯罪的地理变化与社会人口统计决定因素之间的关联。这些分析基于从尼日利亚国家统计局获得的2017年报告的犯罪统计数据。本文分析了四种犯罪类型（武装抢劫，盗窃，强奸和绑架）与其在尼日利亚各州的地理分布有关的空间格局。与传统的回归分析相比，泊松混合模型的制定是考虑了空间依赖性效应（聚类）和特定的国家级犯罪异质性效应。该研究建立了六个解释变量的模型（失业率，人口密度，教育指数，国民总收入（GNI），尼日利亚人口中的男性百分比（PMP），年龄18-35岁和警务结构）。通过马尔可夫链蒙特卡洛模拟的完整贝叶斯方法被用来估计模型参数。结果表明，失业率与强奸，绑架和武装抢劫呈正相关，而与盗窃呈负相关。结果进一步表明，GNI和PMP与所有犯罪均呈正相关。除了犯罪的危险因素外，武装抢劫事件中占总变化集聚作用的比例变化还可以解释为29.27％，盗窃（偷窃）事件占31.30％，绑架事件占27.07％，强奸案发生时占41.40％。 。我们的方法还可以生成空间预测图，以识别犯罪高度集中的地区，